OpenCores
URL https://opencores.org/ocsvn/open8_urisc/open8_urisc/trunk

Subversion Repositories open8_urisc

[/] [open8_urisc/] [trunk/] [gnu/] [binutils/] [bfd/] [elf32-arm.c] - Rev 90

Go to most recent revision | Compare with Previous | Blame | View Log

/* 32-bit ELF support for ARM
   Copyright 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007,
   2008, 2009, 2010, 2011  Free Software Foundation, Inc.
 
   This file is part of BFD, the Binary File Descriptor library.
 
   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 3 of the License, or
   (at your option) any later version.
 
   This program is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU General Public License for more details.
 
   You should have received a copy of the GNU General Public License
   along with this program; if not, write to the Free Software
   Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston,
   MA 02110-1301, USA.  */
 
#include "sysdep.h"
#include <limits.h>
 
#include "bfd.h"
#include "libiberty.h"
#include "libbfd.h"
#include "elf-bfd.h"
#include "elf-vxworks.h"
#include "elf/arm.h"
 
/* Return the relocation section associated with NAME.  HTAB is the
   bfd's elf32_arm_link_hash_entry.  */
#define RELOC_SECTION(HTAB, NAME) \
  ((HTAB)->use_rel ? ".rel" NAME : ".rela" NAME)
 
/* Return size of a relocation entry.  HTAB is the bfd's
   elf32_arm_link_hash_entry.  */
#define RELOC_SIZE(HTAB) \
  ((HTAB)->use_rel \
   ? sizeof (Elf32_External_Rel) \
   : sizeof (Elf32_External_Rela))
 
/* Return function to swap relocations in.  HTAB is the bfd's
   elf32_arm_link_hash_entry.  */
#define SWAP_RELOC_IN(HTAB) \
  ((HTAB)->use_rel \
   ? bfd_elf32_swap_reloc_in \
   : bfd_elf32_swap_reloca_in)
 
/* Return function to swap relocations out.  HTAB is the bfd's
   elf32_arm_link_hash_entry.  */
#define SWAP_RELOC_OUT(HTAB) \
  ((HTAB)->use_rel \
   ? bfd_elf32_swap_reloc_out \
   : bfd_elf32_swap_reloca_out)
 
#define elf_info_to_howto               0
#define elf_info_to_howto_rel           elf32_arm_info_to_howto
 
#define ARM_ELF_ABI_VERSION		0
#define ARM_ELF_OS_ABI_VERSION		ELFOSABI_ARM
 
static bfd_boolean elf32_arm_write_section (bfd *output_bfd,
					    struct bfd_link_info *link_info,
					    asection *sec,
					    bfd_byte *contents);
 
/* Note: code such as elf32_arm_reloc_type_lookup expect to use e.g.
   R_ARM_PC24 as an index into this, and find the R_ARM_PC24 HOWTO
   in that slot.  */
 
static reloc_howto_type elf32_arm_howto_table_1[] =
{
  /* No relocation.  */
  HOWTO (R_ARM_NONE,		/* type */
	 0,			/* rightshift */
	 0,			/* size (0 = byte, 1 = short, 2 = long) */
	 0,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_NONE",		/* name */
	 FALSE,			/* partial_inplace */
	 0,			/* src_mask */
	 0,			/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_PC24,		/* type */
	 2,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 24,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_PC24",		/* name */
	 FALSE,			/* partial_inplace */
	 0x00ffffff,		/* src_mask */
	 0x00ffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  /* 32 bit absolute */
  HOWTO (R_ARM_ABS32,		/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ABS32",		/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  /* standard 32bit pc-relative reloc */
  HOWTO (R_ARM_REL32,		/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_REL32",		/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  /* 8 bit absolute - R_ARM_LDR_PC_G0 in AAELF */
  HOWTO (R_ARM_LDR_PC_G0,	/* type */
	 0,			/* rightshift */
	 0,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDR_PC_G0",     /* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
   /* 16 bit absolute */
  HOWTO (R_ARM_ABS16,		/* type */
	 0,			/* rightshift */
	 1,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ABS16",		/* name */
	 FALSE,			/* partial_inplace */
	 0x0000ffff,		/* src_mask */
	 0x0000ffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  /* 12 bit absolute */
  HOWTO (R_ARM_ABS12,		/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 12,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ABS12",		/* name */
	 FALSE,			/* partial_inplace */
	 0x00000fff,		/* src_mask */
	 0x00000fff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_THM_ABS5,	/* type */
	 6,			/* rightshift */
	 1,			/* size (0 = byte, 1 = short, 2 = long) */
	 5,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_ABS5",	/* name */
	 FALSE,			/* partial_inplace */
	 0x000007e0,		/* src_mask */
	 0x000007e0,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  /* 8 bit absolute */
  HOWTO (R_ARM_ABS8,		/* type */
	 0,			/* rightshift */
	 0,			/* size (0 = byte, 1 = short, 2 = long) */
	 8,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ABS8",		/* name */
	 FALSE,			/* partial_inplace */
	 0x000000ff,		/* src_mask */
	 0x000000ff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_SBREL32,		/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_SBREL32",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_THM_CALL,	/* type */
	 1,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 24,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_CALL",	/* name */
	 FALSE,			/* partial_inplace */
	 0x07ff07ff,		/* src_mask */
	 0x07ff07ff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_THM_PC8,	        /* type */
	 1,			/* rightshift */
	 1,			/* size (0 = byte, 1 = short, 2 = long) */
	 8,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_PC8",	/* name */
	 FALSE,			/* partial_inplace */
	 0x000000ff,		/* src_mask */
	 0x000000ff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_BREL_ADJ,	/* type */
	 1,			/* rightshift */
	 1,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_BREL_ADJ",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_TLS_DESC,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_TLS_DESC",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_THM_SWI8,	/* type */
	 0,			/* rightshift */
	 0,			/* size (0 = byte, 1 = short, 2 = long) */
	 0,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_SWI8",		/* name */
	 FALSE,			/* partial_inplace */
	 0x00000000,		/* src_mask */
	 0x00000000,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  /* BLX instruction for the ARM.  */
  HOWTO (R_ARM_XPC25,		/* type */
	 2,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 25,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_XPC25",		/* name */
	 FALSE,			/* partial_inplace */
	 0x00ffffff,		/* src_mask */
	 0x00ffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  /* BLX instruction for the Thumb.  */
  HOWTO (R_ARM_THM_XPC22,	/* type */
	 2,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 22,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_XPC22",	/* name */
	 FALSE,			/* partial_inplace */
	 0x07ff07ff,		/* src_mask */
	 0x07ff07ff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  /* Dynamic TLS relocations.  */
 
  HOWTO (R_ARM_TLS_DTPMOD32,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_TLS_DTPMOD32",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_TLS_DTPOFF32,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_TLS_DTPOFF32",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_TLS_TPOFF32,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_TLS_TPOFF32",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  /* Relocs used in ARM Linux */
 
  HOWTO (R_ARM_COPY,		/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_COPY",		/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_GLOB_DAT,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_GLOB_DAT",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_JUMP_SLOT,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_JUMP_SLOT",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_RELATIVE,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_RELATIVE",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_GOTOFF32,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_GOTOFF32",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_GOTPC,		/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         TRUE,			/* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_GOTPC",		/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_GOT32,		/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,			/* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_GOT32",		/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_PLT32,		/* type */
         2,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         24,                    /* bitsize */
         TRUE,			/* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_PLT32",		/* name */
         FALSE,			/* partial_inplace */
         0x00ffffff,		/* src_mask */
         0x00ffffff,		/* dst_mask */
         TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_CALL,		/* type */
	 2,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 24,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_CALL",		/* name */
	 FALSE,			/* partial_inplace */
	 0x00ffffff,		/* src_mask */
	 0x00ffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_JUMP24,		/* type */
	 2,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 24,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_JUMP24",	/* name */
	 FALSE,			/* partial_inplace */
	 0x00ffffff,		/* src_mask */
	 0x00ffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_THM_JUMP24,	/* type */
	 1,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 24,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_JUMP24",	/* name */
	 FALSE,			/* partial_inplace */
	 0x07ff2fff,		/* src_mask */
	 0x07ff2fff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_BASE_ABS,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_BASE_ABS",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_PCREL7_0,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 12,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_PCREL_7_0",	/* name */
	 FALSE,			/* partial_inplace */
	 0x00000fff,		/* src_mask */
	 0x00000fff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_PCREL15_8,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 12,			/* bitsize */
	 TRUE,			/* pc_relative */
	 8,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_PCREL_15_8",/* name */
	 FALSE,			/* partial_inplace */
	 0x00000fff,		/* src_mask */
	 0x00000fff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_PCREL23_15,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 12,			/* bitsize */
	 TRUE,			/* pc_relative */
	 16,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_PCREL_23_15",/* name */
	 FALSE,			/* partial_inplace */
	 0x00000fff,		/* src_mask */
	 0x00000fff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDR_SBREL_11_0,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 12,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDR_SBREL_11_0",/* name */
	 FALSE,			/* partial_inplace */
	 0x00000fff,		/* src_mask */
	 0x00000fff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_SBREL_19_12,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 8,			/* bitsize */
	 FALSE,			/* pc_relative */
	 12,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_SBREL_19_12",/* name */
	 FALSE,			/* partial_inplace */
	 0x000ff000,		/* src_mask */
	 0x000ff000,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_SBREL_27_20,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 8,			/* bitsize */
	 FALSE,			/* pc_relative */
	 20,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_SBREL_27_20",/* name */
	 FALSE,			/* partial_inplace */
	 0x0ff00000,		/* src_mask */
	 0x0ff00000,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_TARGET1,		/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_TARGET1",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_ROSEGREL32,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ROSEGREL32",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_V4BX,		/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_V4BX",		/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_TARGET2,		/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_TARGET2",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_PREL31,		/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 31,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_PREL31",	/* name */
	 FALSE,			/* partial_inplace */
	 0x7fffffff,		/* src_mask */
	 0x7fffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_MOVW_ABS_NC,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_MOVW_ABS_NC",	/* name */
	 FALSE,			/* partial_inplace */
	 0x000f0fff,		/* src_mask */
	 0x000f0fff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_MOVT_ABS,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_MOVT_ABS",	/* name */
	 FALSE,			/* partial_inplace */
	 0x000f0fff,		/* src_mask */
	 0x000f0fff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_MOVW_PREL_NC,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_MOVW_PREL_NC",	/* name */
	 FALSE,			/* partial_inplace */
	 0x000f0fff,		/* src_mask */
	 0x000f0fff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_MOVT_PREL,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_MOVT_PREL",	/* name */
	 FALSE,			/* partial_inplace */
	 0x000f0fff,		/* src_mask */
	 0x000f0fff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_THM_MOVW_ABS_NC,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_MOVW_ABS_NC",/* name */
	 FALSE,			/* partial_inplace */
	 0x040f70ff,		/* src_mask */
	 0x040f70ff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_THM_MOVT_ABS,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_MOVT_ABS",	/* name */
	 FALSE,			/* partial_inplace */
	 0x040f70ff,		/* src_mask */
	 0x040f70ff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_THM_MOVW_PREL_NC,/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_MOVW_PREL_NC",/* name */
	 FALSE,			/* partial_inplace */
	 0x040f70ff,		/* src_mask */
	 0x040f70ff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_THM_MOVT_PREL,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_MOVT_PREL",	/* name */
	 FALSE,			/* partial_inplace */
	 0x040f70ff,		/* src_mask */
	 0x040f70ff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_THM_JUMP19,	/* type */
	 1,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 19,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,/* complain_on_overflow */
	 bfd_elf_generic_reloc, /* special_function */
	 "R_ARM_THM_JUMP19",	/* name */
	 FALSE,			/* partial_inplace */
	 0x043f2fff,		/* src_mask */
	 0x043f2fff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_THM_JUMP6,	/* type */
	 1,			/* rightshift */
	 1,			/* size (0 = byte, 1 = short, 2 = long) */
	 6,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_unsigned,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_JUMP6",	/* name */
	 FALSE,			/* partial_inplace */
	 0x02f8,		/* src_mask */
	 0x02f8,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  /* These are declared as 13-bit signed relocations because we can
     address -4095 .. 4095(base) by altering ADDW to SUBW or vice
     versa.  */
  HOWTO (R_ARM_THM_ALU_PREL_11_0,/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 13,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_ALU_PREL_11_0",/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_THM_PC12,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 13,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_PC12",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_ABS32_NOI,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ABS32_NOI",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_REL32_NOI,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_REL32_NOI",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  /* Group relocations.  */
 
  HOWTO (R_ARM_ALU_PC_G0_NC,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_PC_G0_NC",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_PC_G0,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_PC_G0",   	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_PC_G1_NC,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_PC_G1_NC",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_PC_G1,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_PC_G1",   	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_PC_G2,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_PC_G2",   	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDR_PC_G1,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDR_PC_G1",   	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDR_PC_G2,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDR_PC_G2",   	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDRS_PC_G0,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDRS_PC_G0",   	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDRS_PC_G1,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDRS_PC_G1",   	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDRS_PC_G2,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDRS_PC_G2",   	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDC_PC_G0,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDC_PC_G0",   	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDC_PC_G1,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDC_PC_G1",   	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDC_PC_G2,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDC_PC_G2",   	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_SB_G0_NC,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_SB_G0_NC", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_SB_G0,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_SB_G0", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_SB_G1_NC,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_SB_G1_NC", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_SB_G1,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_SB_G1", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_ALU_SB_G2,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_ALU_SB_G2", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDR_SB_G0,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDR_SB_G0", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDR_SB_G1,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDR_SB_G1", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDR_SB_G2,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDR_SB_G2", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDRS_SB_G0,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDRS_SB_G0", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDRS_SB_G1,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDRS_SB_G1", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDRS_SB_G2,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDRS_SB_G2", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDC_SB_G0,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDC_SB_G0", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDC_SB_G1,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDC_SB_G1", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_LDC_SB_G2,   	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_LDC_SB_G2", 	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  /* End of group relocations.  */
 
  HOWTO (R_ARM_MOVW_BREL_NC,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_MOVW_BREL_NC",	/* name */
	 FALSE,			/* partial_inplace */
	 0x0000ffff,		/* src_mask */
	 0x0000ffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_MOVT_BREL,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_MOVT_BREL",	/* name */
	 FALSE,			/* partial_inplace */
	 0x0000ffff,		/* src_mask */
	 0x0000ffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_MOVW_BREL,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_MOVW_BREL",	/* name */
	 FALSE,			/* partial_inplace */
	 0x0000ffff,		/* src_mask */
	 0x0000ffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_THM_MOVW_BREL_NC,/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_MOVW_BREL_NC",/* name */
	 FALSE,			/* partial_inplace */
	 0x040f70ff,		/* src_mask */
	 0x040f70ff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_THM_MOVT_BREL,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_MOVT_BREL",	/* name */
	 FALSE,			/* partial_inplace */
	 0x040f70ff,		/* src_mask */
	 0x040f70ff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_THM_MOVW_BREL,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 16,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_MOVW_BREL",	/* name */
	 FALSE,			/* partial_inplace */
	 0x040f70ff,		/* src_mask */
	 0x040f70ff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_TLS_GOTDESC,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 NULL,			/* special_function */
	 "R_ARM_TLS_GOTDESC",	/* name */
	 TRUE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_TLS_CALL,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 24,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_TLS_CALL",	/* name */
	 FALSE,			/* partial_inplace */
	 0x00ffffff,		/* src_mask */
	 0x00ffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_TLS_DESCSEQ,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 0,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_TLS_DESCSEQ",	/* name */
	 FALSE,			/* partial_inplace */
	 0x00000000,		/* src_mask */
	 0x00000000,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_THM_TLS_CALL,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 24,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_TLS_CALL",	/* name */
	 FALSE,			/* partial_inplace */
	 0x07ff07ff,		/* src_mask */
	 0x07ff07ff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_PLT32_ABS,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_PLT32_ABS",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_GOT_ABS,		/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_GOT_ABS",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 FALSE),			/* pcrel_offset */
 
  HOWTO (R_ARM_GOT_PREL,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 32,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,	/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_GOT_PREL",	/* name */
	 FALSE,			/* partial_inplace */
	 0xffffffff,		/* src_mask */
	 0xffffffff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_GOT_BREL12,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 12,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_GOT_BREL12",	/* name */
	 FALSE,			/* partial_inplace */
	 0x00000fff,		/* src_mask */
	 0x00000fff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_GOTOFF12,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 12,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_GOTOFF12",	/* name */
	 FALSE,			/* partial_inplace */
	 0x00000fff,		/* src_mask */
	 0x00000fff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  EMPTY_HOWTO (R_ARM_GOTRELAX),  /* reserved for future GOT-load optimizations */
 
  /* GNU extension to record C++ vtable member usage */
  HOWTO (R_ARM_GNU_VTENTRY,     /* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         0,                     /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_dont, /* complain_on_overflow */
         _bfd_elf_rel_vtable_reloc_fn,  /* special_function */
         "R_ARM_GNU_VTENTRY",   /* name */
         FALSE,                 /* partial_inplace */
         0,                     /* src_mask */
         0,                     /* dst_mask */
         FALSE),                /* pcrel_offset */
 
  /* GNU extension to record C++ vtable hierarchy */
  HOWTO (R_ARM_GNU_VTINHERIT, /* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         0,                     /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_dont, /* complain_on_overflow */
         NULL,                  /* special_function */
         "R_ARM_GNU_VTINHERIT", /* name */
         FALSE,                 /* partial_inplace */
         0,                     /* src_mask */
         0,                     /* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_THM_JUMP11,	/* type */
	 1,			/* rightshift */
	 1,			/* size (0 = byte, 1 = short, 2 = long) */
	 11,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,	/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_JUMP11",	/* name */
	 FALSE,			/* partial_inplace */
	 0x000007ff,		/* src_mask */
	 0x000007ff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  HOWTO (R_ARM_THM_JUMP8,	/* type */
	 1,			/* rightshift */
	 1,			/* size (0 = byte, 1 = short, 2 = long) */
	 8,			/* bitsize */
	 TRUE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_signed,	/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_JUMP8",	/* name */
	 FALSE,			/* partial_inplace */
	 0x000000ff,		/* src_mask */
	 0x000000ff,		/* dst_mask */
	 TRUE),			/* pcrel_offset */
 
  /* TLS relocations */
  HOWTO (R_ARM_TLS_GD32,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         NULL,			/* special_function */
         "R_ARM_TLS_GD32",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_TLS_LDM32,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_TLS_LDM32",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_TLS_LDO32,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_TLS_LDO32",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_TLS_IE32,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                  /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         NULL,			/* special_function */
         "R_ARM_TLS_IE32",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_TLS_LE32,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_TLS_LE32",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE),                /* pcrel_offset */
 
  HOWTO (R_ARM_TLS_LDO12,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 12,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_TLS_LDO12",	/* name */
	 FALSE,			/* partial_inplace */
	 0x00000fff,		/* src_mask */
	 0x00000fff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_TLS_LE12,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 12,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_TLS_LE12",	/* name */
	 FALSE,			/* partial_inplace */
	 0x00000fff,		/* src_mask */
	 0x00000fff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_TLS_IE12GP,	/* type */
	 0,			/* rightshift */
	 2,			/* size (0 = byte, 1 = short, 2 = long) */
	 12,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_TLS_IE12GP",	/* name */
	 FALSE,			/* partial_inplace */
	 0x00000fff,		/* src_mask */
	 0x00000fff,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  /* 112-127 private relocations.  */
  EMPTY_HOWTO (112),
  EMPTY_HOWTO (113),
  EMPTY_HOWTO (114),
  EMPTY_HOWTO (115),
  EMPTY_HOWTO (116),
  EMPTY_HOWTO (117),
  EMPTY_HOWTO (118),
  EMPTY_HOWTO (119),
  EMPTY_HOWTO (120),
  EMPTY_HOWTO (121),
  EMPTY_HOWTO (122),
  EMPTY_HOWTO (123),
  EMPTY_HOWTO (124),
  EMPTY_HOWTO (125),
  EMPTY_HOWTO (126),
  EMPTY_HOWTO (127),
 
  /* R_ARM_ME_TOO, obsolete.  */
  EMPTY_HOWTO (128),
 
  HOWTO (R_ARM_THM_TLS_DESCSEQ,	/* type */
	 0,			/* rightshift */
	 1,			/* size (0 = byte, 1 = short, 2 = long) */
	 0,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_bitfield,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_THM_TLS_DESCSEQ",/* name */
	 FALSE,			/* partial_inplace */
	 0x00000000,		/* src_mask */
	 0x00000000,		/* dst_mask */
	 FALSE),		/* pcrel_offset */
};
 
/* 160 onwards: */
static reloc_howto_type elf32_arm_howto_table_2[1] =
{
  HOWTO (R_ARM_IRELATIVE,	/* type */
         0,                     /* rightshift */
         2,                     /* size (0 = byte, 1 = short, 2 = long) */
         32,                    /* bitsize */
         FALSE,                 /* pc_relative */
         0,                     /* bitpos */
         complain_overflow_bitfield,/* complain_on_overflow */
         bfd_elf_generic_reloc, /* special_function */
         "R_ARM_IRELATIVE",	/* name */
         TRUE,			/* partial_inplace */
         0xffffffff,		/* src_mask */
         0xffffffff,		/* dst_mask */
         FALSE)			/* pcrel_offset */
};
 
/* 249-255 extended, currently unused, relocations:  */
static reloc_howto_type elf32_arm_howto_table_3[4] =
{
  HOWTO (R_ARM_RREL32,		/* type */
	 0,			/* rightshift */
	 0,			/* size (0 = byte, 1 = short, 2 = long) */
	 0,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_RREL32",	/* name */
	 FALSE,			/* partial_inplace */
	 0,			/* src_mask */
	 0,			/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_RABS32,		/* type */
	 0,			/* rightshift */
	 0,			/* size (0 = byte, 1 = short, 2 = long) */
	 0,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_RABS32",	/* name */
	 FALSE,			/* partial_inplace */
	 0,			/* src_mask */
	 0,			/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_RPC24,		/* type */
	 0,			/* rightshift */
	 0,			/* size (0 = byte, 1 = short, 2 = long) */
	 0,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_RPC24",		/* name */
	 FALSE,			/* partial_inplace */
	 0,			/* src_mask */
	 0,			/* dst_mask */
	 FALSE),		/* pcrel_offset */
 
  HOWTO (R_ARM_RBASE,		/* type */
	 0,			/* rightshift */
	 0,			/* size (0 = byte, 1 = short, 2 = long) */
	 0,			/* bitsize */
	 FALSE,			/* pc_relative */
	 0,			/* bitpos */
	 complain_overflow_dont,/* complain_on_overflow */
	 bfd_elf_generic_reloc,	/* special_function */
	 "R_ARM_RBASE",		/* name */
	 FALSE,			/* partial_inplace */
	 0,			/* src_mask */
	 0,			/* dst_mask */
	 FALSE)			/* pcrel_offset */
};
 
static reloc_howto_type *
elf32_arm_howto_from_type (unsigned int r_type)
{
  if (r_type < ARRAY_SIZE (elf32_arm_howto_table_1))
    return &elf32_arm_howto_table_1[r_type];
 
  if (r_type == R_ARM_IRELATIVE)
    return &elf32_arm_howto_table_2[r_type - R_ARM_IRELATIVE];
 
  if (r_type >= R_ARM_RREL32
      && r_type < R_ARM_RREL32 + ARRAY_SIZE (elf32_arm_howto_table_3))
    return &elf32_arm_howto_table_3[r_type - R_ARM_RREL32];
 
  return NULL;
}
 
static void
elf32_arm_info_to_howto (bfd * abfd ATTRIBUTE_UNUSED, arelent * bfd_reloc,
			 Elf_Internal_Rela * elf_reloc)
{
  unsigned int r_type;
 
  r_type = ELF32_R_TYPE (elf_reloc->r_info);
  bfd_reloc->howto = elf32_arm_howto_from_type (r_type);
}
 
struct elf32_arm_reloc_map
  {
    bfd_reloc_code_real_type  bfd_reloc_val;
    unsigned char             elf_reloc_val;
  };
 
/* All entries in this list must also be present in elf32_arm_howto_table.  */
static const struct elf32_arm_reloc_map elf32_arm_reloc_map[] =
  {
    {BFD_RELOC_NONE,                 R_ARM_NONE},
    {BFD_RELOC_ARM_PCREL_BRANCH,     R_ARM_PC24},
    {BFD_RELOC_ARM_PCREL_CALL,	     R_ARM_CALL},
    {BFD_RELOC_ARM_PCREL_JUMP,	     R_ARM_JUMP24},
    {BFD_RELOC_ARM_PCREL_BLX,        R_ARM_XPC25},
    {BFD_RELOC_THUMB_PCREL_BLX,      R_ARM_THM_XPC22},
    {BFD_RELOC_32,                   R_ARM_ABS32},
    {BFD_RELOC_32_PCREL,             R_ARM_REL32},
    {BFD_RELOC_8,                    R_ARM_ABS8},
    {BFD_RELOC_16,                   R_ARM_ABS16},
    {BFD_RELOC_ARM_OFFSET_IMM,       R_ARM_ABS12},
    {BFD_RELOC_ARM_THUMB_OFFSET,     R_ARM_THM_ABS5},
    {BFD_RELOC_THUMB_PCREL_BRANCH25, R_ARM_THM_JUMP24},
    {BFD_RELOC_THUMB_PCREL_BRANCH23, R_ARM_THM_CALL},
    {BFD_RELOC_THUMB_PCREL_BRANCH12, R_ARM_THM_JUMP11},
    {BFD_RELOC_THUMB_PCREL_BRANCH20, R_ARM_THM_JUMP19},
    {BFD_RELOC_THUMB_PCREL_BRANCH9,  R_ARM_THM_JUMP8},
    {BFD_RELOC_THUMB_PCREL_BRANCH7,  R_ARM_THM_JUMP6},
    {BFD_RELOC_ARM_GLOB_DAT,         R_ARM_GLOB_DAT},
    {BFD_RELOC_ARM_JUMP_SLOT,        R_ARM_JUMP_SLOT},
    {BFD_RELOC_ARM_RELATIVE,         R_ARM_RELATIVE},
    {BFD_RELOC_ARM_GOTOFF,           R_ARM_GOTOFF32},
    {BFD_RELOC_ARM_GOTPC,            R_ARM_GOTPC},
    {BFD_RELOC_ARM_GOT_PREL,         R_ARM_GOT_PREL},
    {BFD_RELOC_ARM_GOT32,            R_ARM_GOT32},
    {BFD_RELOC_ARM_PLT32,            R_ARM_PLT32},
    {BFD_RELOC_ARM_TARGET1,	     R_ARM_TARGET1},
    {BFD_RELOC_ARM_ROSEGREL32,	     R_ARM_ROSEGREL32},
    {BFD_RELOC_ARM_SBREL32,	     R_ARM_SBREL32},
    {BFD_RELOC_ARM_PREL31,	     R_ARM_PREL31},
    {BFD_RELOC_ARM_TARGET2,	     R_ARM_TARGET2},
    {BFD_RELOC_ARM_PLT32,            R_ARM_PLT32},
    {BFD_RELOC_ARM_TLS_GOTDESC,      R_ARM_TLS_GOTDESC},
    {BFD_RELOC_ARM_TLS_CALL,         R_ARM_TLS_CALL},
    {BFD_RELOC_ARM_THM_TLS_CALL,     R_ARM_THM_TLS_CALL},
    {BFD_RELOC_ARM_TLS_DESCSEQ,      R_ARM_TLS_DESCSEQ},
    {BFD_RELOC_ARM_THM_TLS_DESCSEQ,  R_ARM_THM_TLS_DESCSEQ},
    {BFD_RELOC_ARM_TLS_DESC,         R_ARM_TLS_DESC},
    {BFD_RELOC_ARM_TLS_GD32,	     R_ARM_TLS_GD32},
    {BFD_RELOC_ARM_TLS_LDO32,	     R_ARM_TLS_LDO32},
    {BFD_RELOC_ARM_TLS_LDM32,	     R_ARM_TLS_LDM32},
    {BFD_RELOC_ARM_TLS_DTPMOD32,     R_ARM_TLS_DTPMOD32},
    {BFD_RELOC_ARM_TLS_DTPOFF32,     R_ARM_TLS_DTPOFF32},
    {BFD_RELOC_ARM_TLS_TPOFF32,      R_ARM_TLS_TPOFF32},
    {BFD_RELOC_ARM_TLS_IE32,         R_ARM_TLS_IE32},
    {BFD_RELOC_ARM_TLS_LE32,         R_ARM_TLS_LE32},
    {BFD_RELOC_ARM_IRELATIVE,        R_ARM_IRELATIVE},
    {BFD_RELOC_VTABLE_INHERIT,	     R_ARM_GNU_VTINHERIT},
    {BFD_RELOC_VTABLE_ENTRY,	     R_ARM_GNU_VTENTRY},
    {BFD_RELOC_ARM_MOVW,	     R_ARM_MOVW_ABS_NC},
    {BFD_RELOC_ARM_MOVT,	     R_ARM_MOVT_ABS},
    {BFD_RELOC_ARM_MOVW_PCREL,	     R_ARM_MOVW_PREL_NC},
    {BFD_RELOC_ARM_MOVT_PCREL,	     R_ARM_MOVT_PREL},
    {BFD_RELOC_ARM_THUMB_MOVW,	     R_ARM_THM_MOVW_ABS_NC},
    {BFD_RELOC_ARM_THUMB_MOVT,	     R_ARM_THM_MOVT_ABS},
    {BFD_RELOC_ARM_THUMB_MOVW_PCREL, R_ARM_THM_MOVW_PREL_NC},
    {BFD_RELOC_ARM_THUMB_MOVT_PCREL, R_ARM_THM_MOVT_PREL},
    {BFD_RELOC_ARM_ALU_PC_G0_NC, R_ARM_ALU_PC_G0_NC},
    {BFD_RELOC_ARM_ALU_PC_G0, R_ARM_ALU_PC_G0},
    {BFD_RELOC_ARM_ALU_PC_G1_NC, R_ARM_ALU_PC_G1_NC},
    {BFD_RELOC_ARM_ALU_PC_G1, R_ARM_ALU_PC_G1},
    {BFD_RELOC_ARM_ALU_PC_G2, R_ARM_ALU_PC_G2},
    {BFD_RELOC_ARM_LDR_PC_G0, R_ARM_LDR_PC_G0},
    {BFD_RELOC_ARM_LDR_PC_G1, R_ARM_LDR_PC_G1},
    {BFD_RELOC_ARM_LDR_PC_G2, R_ARM_LDR_PC_G2},
    {BFD_RELOC_ARM_LDRS_PC_G0, R_ARM_LDRS_PC_G0},
    {BFD_RELOC_ARM_LDRS_PC_G1, R_ARM_LDRS_PC_G1},
    {BFD_RELOC_ARM_LDRS_PC_G2, R_ARM_LDRS_PC_G2},
    {BFD_RELOC_ARM_LDC_PC_G0, R_ARM_LDC_PC_G0},
    {BFD_RELOC_ARM_LDC_PC_G1, R_ARM_LDC_PC_G1},
    {BFD_RELOC_ARM_LDC_PC_G2, R_ARM_LDC_PC_G2},
    {BFD_RELOC_ARM_ALU_SB_G0_NC, R_ARM_ALU_SB_G0_NC},
    {BFD_RELOC_ARM_ALU_SB_G0, R_ARM_ALU_SB_G0},
    {BFD_RELOC_ARM_ALU_SB_G1_NC, R_ARM_ALU_SB_G1_NC},
    {BFD_RELOC_ARM_ALU_SB_G1, R_ARM_ALU_SB_G1},
    {BFD_RELOC_ARM_ALU_SB_G2, R_ARM_ALU_SB_G2},
    {BFD_RELOC_ARM_LDR_SB_G0, R_ARM_LDR_SB_G0},
    {BFD_RELOC_ARM_LDR_SB_G1, R_ARM_LDR_SB_G1},
    {BFD_RELOC_ARM_LDR_SB_G2, R_ARM_LDR_SB_G2},
    {BFD_RELOC_ARM_LDRS_SB_G0, R_ARM_LDRS_SB_G0},
    {BFD_RELOC_ARM_LDRS_SB_G1, R_ARM_LDRS_SB_G1},
    {BFD_RELOC_ARM_LDRS_SB_G2, R_ARM_LDRS_SB_G2},
    {BFD_RELOC_ARM_LDC_SB_G0, R_ARM_LDC_SB_G0},
    {BFD_RELOC_ARM_LDC_SB_G1, R_ARM_LDC_SB_G1},
    {BFD_RELOC_ARM_LDC_SB_G2, R_ARM_LDC_SB_G2},
    {BFD_RELOC_ARM_V4BX,	     R_ARM_V4BX}
  };
 
static reloc_howto_type *
elf32_arm_reloc_type_lookup (bfd *abfd ATTRIBUTE_UNUSED,
			     bfd_reloc_code_real_type code)
{
  unsigned int i;
 
  for (i = 0; i < ARRAY_SIZE (elf32_arm_reloc_map); i ++)
    if (elf32_arm_reloc_map[i].bfd_reloc_val == code)
      return elf32_arm_howto_from_type (elf32_arm_reloc_map[i].elf_reloc_val);
 
  return NULL;
}
 
static reloc_howto_type *
elf32_arm_reloc_name_lookup (bfd *abfd ATTRIBUTE_UNUSED,
			     const char *r_name)
{
  unsigned int i;
 
  for (i = 0; i < ARRAY_SIZE (elf32_arm_howto_table_1); i++)
    if (elf32_arm_howto_table_1[i].name != NULL
	&& strcasecmp (elf32_arm_howto_table_1[i].name, r_name) == 0)
      return &elf32_arm_howto_table_1[i];
 
  for (i = 0; i < ARRAY_SIZE (elf32_arm_howto_table_2); i++)
    if (elf32_arm_howto_table_2[i].name != NULL
	&& strcasecmp (elf32_arm_howto_table_2[i].name, r_name) == 0)
      return &elf32_arm_howto_table_2[i];
 
  for (i = 0; i < ARRAY_SIZE (elf32_arm_howto_table_3); i++)
    if (elf32_arm_howto_table_3[i].name != NULL
	&& strcasecmp (elf32_arm_howto_table_3[i].name, r_name) == 0)
      return &elf32_arm_howto_table_3[i];
 
  return NULL;
}
 
/* Support for core dump NOTE sections.  */
 
static bfd_boolean
elf32_arm_nabi_grok_prstatus (bfd *abfd, Elf_Internal_Note *note)
{
  int offset;
  size_t size;
 
  switch (note->descsz)
    {
      default:
	return FALSE;
 
      case 148:		/* Linux/ARM 32-bit.  */
	/* pr_cursig */
	elf_tdata (abfd)->core_signal = bfd_get_16 (abfd, note->descdata + 12);
 
	/* pr_pid */
	elf_tdata (abfd)->core_lwpid = bfd_get_32 (abfd, note->descdata + 24);
 
	/* pr_reg */
	offset = 72;
	size = 72;
 
	break;
    }
 
  /* Make a ".reg/999" section.  */
  return _bfd_elfcore_make_pseudosection (abfd, ".reg",
					  size, note->descpos + offset);
}
 
static bfd_boolean
elf32_arm_nabi_grok_psinfo (bfd *abfd, Elf_Internal_Note *note)
{
  switch (note->descsz)
    {
      default:
	return FALSE;
 
      case 124:		/* Linux/ARM elf_prpsinfo.  */
	elf_tdata (abfd)->core_program
	 = _bfd_elfcore_strndup (abfd, note->descdata + 28, 16);
	elf_tdata (abfd)->core_command
	 = _bfd_elfcore_strndup (abfd, note->descdata + 44, 80);
    }
 
  /* Note that for some reason, a spurious space is tacked
     onto the end of the args in some (at least one anyway)
     implementations, so strip it off if it exists.  */
  {
    char *command = elf_tdata (abfd)->core_command;
    int n = strlen (command);
 
    if (0 < n && command[n - 1] == ' ')
      command[n - 1] = '\0';
  }
 
  return TRUE;
}
 
#define TARGET_LITTLE_SYM               bfd_elf32_littlearm_vec
#define TARGET_LITTLE_NAME              "elf32-littlearm"
#define TARGET_BIG_SYM                  bfd_elf32_bigarm_vec
#define TARGET_BIG_NAME                 "elf32-bigarm"
 
#define elf_backend_grok_prstatus	elf32_arm_nabi_grok_prstatus
#define elf_backend_grok_psinfo		elf32_arm_nabi_grok_psinfo
 
typedef unsigned long int insn32;
typedef unsigned short int insn16;
 
/* In lieu of proper flags, assume all EABIv4 or later objects are
   interworkable.  */
#define INTERWORK_FLAG(abfd)  \
  (EF_ARM_EABI_VERSION (elf_elfheader (abfd)->e_flags) >= EF_ARM_EABI_VER4 \
  || (elf_elfheader (abfd)->e_flags & EF_ARM_INTERWORK) \
  || ((abfd)->flags & BFD_LINKER_CREATED))
 
/* The linker script knows the section names for placement.
   The entry_names are used to do simple name mangling on the stubs.
   Given a function name, and its type, the stub can be found. The
   name can be changed. The only requirement is the %s be present.  */
#define THUMB2ARM_GLUE_SECTION_NAME ".glue_7t"
#define THUMB2ARM_GLUE_ENTRY_NAME   "__%s_from_thumb"
 
#define ARM2THUMB_GLUE_SECTION_NAME ".glue_7"
#define ARM2THUMB_GLUE_ENTRY_NAME   "__%s_from_arm"
 
#define VFP11_ERRATUM_VENEER_SECTION_NAME ".vfp11_veneer"
#define VFP11_ERRATUM_VENEER_ENTRY_NAME   "__vfp11_veneer_%x"
 
#define ARM_BX_GLUE_SECTION_NAME ".v4_bx"
#define ARM_BX_GLUE_ENTRY_NAME   "__bx_r%d"
 
#define STUB_ENTRY_NAME   "__%s_veneer"
 
/* The name of the dynamic interpreter.  This is put in the .interp
   section.  */
#define ELF_DYNAMIC_INTERPRETER     "/usr/lib/ld.so.1"
 
static const unsigned long tls_trampoline [] =
  {
    0xe08e0000,		/* add r0, lr, r0 */
    0xe5901004,		/* ldr r1, [r0,#4] */
    0xe12fff11,		/* bx  r1 */
  };
 
static const unsigned long dl_tlsdesc_lazy_trampoline [] =
  {
    0xe52d2004, /*	push    {r2}			*/
    0xe59f200c, /*      ldr     r2, [pc, #3f - . - 8]	*/
    0xe59f100c, /*      ldr     r1, [pc, #4f - . - 8]	*/
    0xe79f2002, /* 1:   ldr     r2, [pc, r2]		*/
    0xe081100f, /* 2:   add     r1, pc			*/
    0xe12fff12, /*      bx      r2			*/
    0x00000014, /* 3:   .word  _GLOBAL_OFFSET_TABLE_ - 1b - 8
		   		+ dl_tlsdesc_lazy_resolver(GOT)   */
    0x00000018, /* 4:   .word  _GLOBAL_OFFSET_TABLE_ - 2b - 8 */ 
  };
 
#ifdef FOUR_WORD_PLT
 
/* The first entry in a procedure linkage table looks like
   this.  It is set up so that any shared library function that is
   called before the relocation has been set up calls the dynamic
   linker first.  */
static const bfd_vma elf32_arm_plt0_entry [] =
  {
    0xe52de004,		/* str   lr, [sp, #-4]! */
    0xe59fe010,		/* ldr   lr, [pc, #16]  */
    0xe08fe00e,		/* add   lr, pc, lr     */
    0xe5bef008,		/* ldr   pc, [lr, #8]!  */
  };
 
/* Subsequent entries in a procedure linkage table look like
   this.  */
static const bfd_vma elf32_arm_plt_entry [] =
  {
    0xe28fc600,		/* add   ip, pc, #NN	*/
    0xe28cca00,		/* add	 ip, ip, #NN	*/
    0xe5bcf000,		/* ldr	 pc, [ip, #NN]! */
    0x00000000,		/* unused		*/
  };
 
#else
 
/* The first entry in a procedure linkage table looks like
   this.  It is set up so that any shared library function that is
   called before the relocation has been set up calls the dynamic
   linker first.  */
static const bfd_vma elf32_arm_plt0_entry [] =
  {
    0xe52de004,		/* str   lr, [sp, #-4]! */
    0xe59fe004,		/* ldr   lr, [pc, #4]   */
    0xe08fe00e,		/* add   lr, pc, lr     */
    0xe5bef008,		/* ldr   pc, [lr, #8]!  */
    0x00000000,		/* &GOT[0] - .          */
  };
 
/* Subsequent entries in a procedure linkage table look like
   this.  */
static const bfd_vma elf32_arm_plt_entry [] =
  {
    0xe28fc600,		/* add   ip, pc, #0xNN00000 */
    0xe28cca00,		/* add	 ip, ip, #0xNN000   */
    0xe5bcf000,		/* ldr	 pc, [ip, #0xNNN]!  */
  };
 
#endif
 
/* The format of the first entry in the procedure linkage table
   for a VxWorks executable.  */
static const bfd_vma elf32_arm_vxworks_exec_plt0_entry[] =
  {
    0xe52dc008,	        /* str    ip,[sp,#-8]!			*/
    0xe59fc000,         /* ldr    ip,[pc]			*/
    0xe59cf008,         /* ldr    pc,[ip,#8]			*/
    0x00000000,         /* .long  _GLOBAL_OFFSET_TABLE_		*/
  };
 
/* The format of subsequent entries in a VxWorks executable.  */
static const bfd_vma elf32_arm_vxworks_exec_plt_entry[] =
  {
    0xe59fc000,         /* ldr    ip,[pc]			*/
    0xe59cf000,         /* ldr    pc,[ip]			*/
    0x00000000,         /* .long  @got				*/
    0xe59fc000,         /* ldr    ip,[pc]			*/
    0xea000000,         /* b      _PLT				*/
    0x00000000,         /* .long  @pltindex*sizeof(Elf32_Rela)	*/
  };
 
/* The format of entries in a VxWorks shared library.  */
static const bfd_vma elf32_arm_vxworks_shared_plt_entry[] =
  {
    0xe59fc000,         /* ldr    ip,[pc]			*/
    0xe79cf009,         /* ldr    pc,[ip,r9]			*/
    0x00000000,         /* .long  @got				*/
    0xe59fc000,         /* ldr    ip,[pc]			*/
    0xe599f008,         /* ldr    pc,[r9,#8]			*/
    0x00000000,         /* .long  @pltindex*sizeof(Elf32_Rela)	*/
  };
 
/* An initial stub used if the PLT entry is referenced from Thumb code.  */
#define PLT_THUMB_STUB_SIZE 4
static const bfd_vma elf32_arm_plt_thumb_stub [] =
  {
    0x4778,		/* bx pc */
    0x46c0		/* nop   */
  };
 
/* The entries in a PLT when using a DLL-based target with multiple
   address spaces.  */
static const bfd_vma elf32_arm_symbian_plt_entry [] =
  {
    0xe51ff004,         /* ldr   pc, [pc, #-4] */
    0x00000000,         /* dcd   R_ARM_GLOB_DAT(X) */
  };
 
#define ARM_MAX_FWD_BRANCH_OFFSET  ((((1 << 23) - 1) << 2) + 8)
#define ARM_MAX_BWD_BRANCH_OFFSET  ((-((1 << 23) << 2)) + 8)
#define THM_MAX_FWD_BRANCH_OFFSET  ((1 << 22) -2 + 4)
#define THM_MAX_BWD_BRANCH_OFFSET  (-(1 << 22) + 4)
#define THM2_MAX_FWD_BRANCH_OFFSET (((1 << 24) - 2) + 4)
#define THM2_MAX_BWD_BRANCH_OFFSET (-(1 << 24) + 4)
 
enum stub_insn_type
  {
    THUMB16_TYPE = 1,
    THUMB32_TYPE,
    ARM_TYPE,
    DATA_TYPE
  };
 
#define THUMB16_INSN(X)		{(X), THUMB16_TYPE, R_ARM_NONE, 0}
/* A bit of a hack.  A Thumb conditional branch, in which the proper condition
   is inserted in arm_build_one_stub().  */
#define THUMB16_BCOND_INSN(X)	{(X), THUMB16_TYPE, R_ARM_NONE, 1}
#define THUMB32_INSN(X)		{(X), THUMB32_TYPE, R_ARM_NONE, 0}
#define THUMB32_B_INSN(X, Z)	{(X), THUMB32_TYPE, R_ARM_THM_JUMP24, (Z)}
#define ARM_INSN(X)		{(X), ARM_TYPE, R_ARM_NONE, 0}
#define ARM_REL_INSN(X, Z)	{(X), ARM_TYPE, R_ARM_JUMP24, (Z)}
#define DATA_WORD(X,Y,Z)	{(X), DATA_TYPE, (Y), (Z)}
 
typedef struct
{
  bfd_vma data;
  enum stub_insn_type type;
  unsigned int r_type;
  int reloc_addend;
}  insn_sequence;
 
/* Arm/Thumb -> Arm/Thumb long branch stub. On V5T and above, use blx
   to reach the stub if necessary.  */
static const insn_sequence elf32_arm_stub_long_branch_any_any[] =
  {
    ARM_INSN(0xe51ff004),            /* ldr   pc, [pc, #-4] */
    DATA_WORD(0, R_ARM_ABS32, 0),    /* dcd   R_ARM_ABS32(X) */
  };
 
/* V4T Arm -> Thumb long branch stub. Used on V4T where blx is not
   available.  */
static const insn_sequence elf32_arm_stub_long_branch_v4t_arm_thumb[] =
  {
    ARM_INSN(0xe59fc000),            /* ldr   ip, [pc, #0] */
    ARM_INSN(0xe12fff1c),            /* bx    ip */
    DATA_WORD(0, R_ARM_ABS32, 0),    /* dcd   R_ARM_ABS32(X) */
  };
 
/* Thumb -> Thumb long branch stub. Used on M-profile architectures.  */
static const insn_sequence elf32_arm_stub_long_branch_thumb_only[] =
  {
    THUMB16_INSN(0xb401),             /* push {r0} */
    THUMB16_INSN(0x4802),             /* ldr  r0, [pc, #8] */
    THUMB16_INSN(0x4684),             /* mov  ip, r0 */
    THUMB16_INSN(0xbc01),             /* pop  {r0} */
    THUMB16_INSN(0x4760),             /* bx   ip */
    THUMB16_INSN(0xbf00),             /* nop */
    DATA_WORD(0, R_ARM_ABS32, 0),     /* dcd  R_ARM_ABS32(X) */
  };
 
/* V4T Thumb -> Thumb long branch stub. Using the stack is not
   allowed.  */
static const insn_sequence elf32_arm_stub_long_branch_v4t_thumb_thumb[] =
  {
    THUMB16_INSN(0x4778),             /* bx   pc */
    THUMB16_INSN(0x46c0),             /* nop */
    ARM_INSN(0xe59fc000),             /* ldr  ip, [pc, #0] */
    ARM_INSN(0xe12fff1c),             /* bx   ip */
    DATA_WORD(0, R_ARM_ABS32, 0),     /* dcd  R_ARM_ABS32(X) */
  };
 
/* V4T Thumb -> ARM long branch stub. Used on V4T where blx is not
   available.  */
static const insn_sequence elf32_arm_stub_long_branch_v4t_thumb_arm[] =
  {
    THUMB16_INSN(0x4778),             /* bx   pc */
    THUMB16_INSN(0x46c0),             /* nop   */
    ARM_INSN(0xe51ff004),             /* ldr   pc, [pc, #-4] */
    DATA_WORD(0, R_ARM_ABS32, 0),     /* dcd   R_ARM_ABS32(X) */
  };
 
/* V4T Thumb -> ARM short branch stub. Shorter variant of the above
   one, when the destination is close enough.  */
static const insn_sequence elf32_arm_stub_short_branch_v4t_thumb_arm[] =
  {
    THUMB16_INSN(0x4778),             /* bx   pc */
    THUMB16_INSN(0x46c0),             /* nop   */
    ARM_REL_INSN(0xea000000, -8),     /* b    (X-8) */
  };
 
/* ARM/Thumb -> ARM long branch stub, PIC.  On V5T and above, use
   blx to reach the stub if necessary.  */
static const insn_sequence elf32_arm_stub_long_branch_any_arm_pic[] =
  {
    ARM_INSN(0xe59fc000),             /* ldr   ip, [pc] */
    ARM_INSN(0xe08ff00c),             /* add   pc, pc, ip */
    DATA_WORD(0, R_ARM_REL32, -4),    /* dcd   R_ARM_REL32(X-4) */
  };
 
/* ARM/Thumb -> Thumb long branch stub, PIC.  On V5T and above, use
   blx to reach the stub if necessary.  We can not add into pc;
   it is not guaranteed to mode switch (different in ARMv6 and
   ARMv7).  */
static const insn_sequence elf32_arm_stub_long_branch_any_thumb_pic[] =
  {
    ARM_INSN(0xe59fc004),             /* ldr   ip, [pc, #4] */
    ARM_INSN(0xe08fc00c),             /* add   ip, pc, ip */
    ARM_INSN(0xe12fff1c),             /* bx    ip */
    DATA_WORD(0, R_ARM_REL32, 0),     /* dcd   R_ARM_REL32(X) */
  };
 
/* V4T ARM -> ARM long branch stub, PIC.  */
static const insn_sequence elf32_arm_stub_long_branch_v4t_arm_thumb_pic[] =
  {
    ARM_INSN(0xe59fc004),             /* ldr   ip, [pc, #4] */
    ARM_INSN(0xe08fc00c),             /* add   ip, pc, ip */
    ARM_INSN(0xe12fff1c),             /* bx    ip */
    DATA_WORD(0, R_ARM_REL32, 0),     /* dcd   R_ARM_REL32(X) */
  };
 
/* V4T Thumb -> ARM long branch stub, PIC.  */
static const insn_sequence elf32_arm_stub_long_branch_v4t_thumb_arm_pic[] =
  {
    THUMB16_INSN(0x4778),             /* bx   pc */
    THUMB16_INSN(0x46c0),             /* nop  */
    ARM_INSN(0xe59fc000),             /* ldr  ip, [pc, #0] */
    ARM_INSN(0xe08cf00f),             /* add  pc, ip, pc */
    DATA_WORD(0, R_ARM_REL32, -4),     /* dcd  R_ARM_REL32(X) */
  };
 
/* Thumb -> Thumb long branch stub, PIC. Used on M-profile
   architectures.  */
static const insn_sequence elf32_arm_stub_long_branch_thumb_only_pic[] =
  {
    THUMB16_INSN(0xb401),             /* push {r0} */
    THUMB16_INSN(0x4802),             /* ldr  r0, [pc, #8] */
    THUMB16_INSN(0x46fc),             /* mov  ip, pc */
    THUMB16_INSN(0x4484),             /* add  ip, r0 */
    THUMB16_INSN(0xbc01),             /* pop  {r0} */
    THUMB16_INSN(0x4760),             /* bx   ip */
    DATA_WORD(0, R_ARM_REL32, 4),     /* dcd  R_ARM_REL32(X) */
  };
 
/* V4T Thumb -> Thumb long branch stub, PIC. Using the stack is not
   allowed.  */
static const insn_sequence elf32_arm_stub_long_branch_v4t_thumb_thumb_pic[] =
  {
    THUMB16_INSN(0x4778),             /* bx   pc */
    THUMB16_INSN(0x46c0),             /* nop */
    ARM_INSN(0xe59fc004),             /* ldr  ip, [pc, #4] */
    ARM_INSN(0xe08fc00c),             /* add   ip, pc, ip */
    ARM_INSN(0xe12fff1c),             /* bx   ip */
    DATA_WORD(0, R_ARM_REL32, 0),     /* dcd  R_ARM_REL32(X) */
  };
 
/* Thumb2/ARM -> TLS trampoline.  Lowest common denominator, which is a
   long PIC stub.  We can use r1 as a scratch -- and cannot use ip.  */
static const insn_sequence elf32_arm_stub_long_branch_any_tls_pic[] =
{
    ARM_INSN(0xe59f1000),             /* ldr   r1, [pc] */
    ARM_INSN(0xe08ff001),             /* add   pc, pc, r1 */
    DATA_WORD(0, R_ARM_REL32, -4),    /* dcd   R_ARM_REL32(X-4) */
};
 
/* V4T Thumb -> TLS trampoline.  lowest common denominator, which is a
   long PIC stub.  We can use r1 as a scratch -- and cannot use ip.  */
static const insn_sequence elf32_arm_stub_long_branch_v4t_thumb_tls_pic[] =
{
    THUMB16_INSN(0x4778),             /* bx   pc */
    THUMB16_INSN(0x46c0),             /* nop */
    ARM_INSN(0xe59f1000),             /* ldr  r1, [pc, #0] */
    ARM_INSN(0xe081f00f),             /* add  pc, r1, pc */
    DATA_WORD(0, R_ARM_REL32, -4),    /* dcd  R_ARM_REL32(X) */
};
 
/* Cortex-A8 erratum-workaround stubs.  */
 
/* Stub used for conditional branches (which may be beyond +/-1MB away, so we
   can't use a conditional branch to reach this stub).  */
 
static const insn_sequence elf32_arm_stub_a8_veneer_b_cond[] =
  {
    THUMB16_BCOND_INSN(0xd001),         /* b<cond>.n true.  */
    THUMB32_B_INSN(0xf000b800, -4),     /* b.w insn_after_original_branch.  */
    THUMB32_B_INSN(0xf000b800, -4)      /* true: b.w original_branch_dest.  */
  };
 
/* Stub used for b.w and bl.w instructions.  */
 
static const insn_sequence elf32_arm_stub_a8_veneer_b[] =
  {
    THUMB32_B_INSN(0xf000b800, -4)	/* b.w original_branch_dest.  */
  };
 
static const insn_sequence elf32_arm_stub_a8_veneer_bl[] =
  {
    THUMB32_B_INSN(0xf000b800, -4)	/* b.w original_branch_dest.  */
  };
 
/* Stub used for Thumb-2 blx.w instructions.  We modified the original blx.w
   instruction (which switches to ARM mode) to point to this stub.  Jump to the
   real destination using an ARM-mode branch.  */
 
static const insn_sequence elf32_arm_stub_a8_veneer_blx[] =
  {
    ARM_REL_INSN(0xea000000, -8)	/* b original_branch_dest.  */
  };
 
/* Section name for stubs is the associated section name plus this
   string.  */
#define STUB_SUFFIX ".stub"
 
/* One entry per long/short branch stub defined above.  */
#define DEF_STUBS \
  DEF_STUB(long_branch_any_any)	\
  DEF_STUB(long_branch_v4t_arm_thumb) \
  DEF_STUB(long_branch_thumb_only) \
  DEF_STUB(long_branch_v4t_thumb_thumb)	\
  DEF_STUB(long_branch_v4t_thumb_arm) \
  DEF_STUB(short_branch_v4t_thumb_arm) \
  DEF_STUB(long_branch_any_arm_pic) \
  DEF_STUB(long_branch_any_thumb_pic) \
  DEF_STUB(long_branch_v4t_thumb_thumb_pic) \
  DEF_STUB(long_branch_v4t_arm_thumb_pic) \
  DEF_STUB(long_branch_v4t_thumb_arm_pic) \
  DEF_STUB(long_branch_thumb_only_pic) \
  DEF_STUB(long_branch_any_tls_pic) \
  DEF_STUB(long_branch_v4t_thumb_tls_pic) \
  DEF_STUB(a8_veneer_b_cond) \
  DEF_STUB(a8_veneer_b) \
  DEF_STUB(a8_veneer_bl) \
  DEF_STUB(a8_veneer_blx)
 
#define DEF_STUB(x) arm_stub_##x,
enum elf32_arm_stub_type {
  arm_stub_none,
  DEF_STUBS
  /* Note the first a8_veneer type */
  arm_stub_a8_veneer_lwm = arm_stub_a8_veneer_b_cond
};
#undef DEF_STUB
 
typedef struct
{
  const insn_sequence* template_sequence;
  int template_size;
} stub_def;
 
#define DEF_STUB(x) {elf32_arm_stub_##x, ARRAY_SIZE(elf32_arm_stub_##x)},
static const stub_def stub_definitions[] = {
  {NULL, 0},
  DEF_STUBS
};
 
struct elf32_arm_stub_hash_entry
{
  /* Base hash table entry structure.  */
  struct bfd_hash_entry root;
 
  /* The stub section.  */
  asection *stub_sec;
 
  /* Offset within stub_sec of the beginning of this stub.  */
  bfd_vma stub_offset;
 
  /* Given the symbol's value and its section we can determine its final
     value when building the stubs (so the stub knows where to jump).  */
  bfd_vma target_value;
  asection *target_section;
 
  /* Offset to apply to relocation referencing target_value.  */
  bfd_vma target_addend;
 
  /* The instruction which caused this stub to be generated (only valid for
     Cortex-A8 erratum workaround stubs at present).  */
  unsigned long orig_insn;
 
  /* The stub type.  */
  enum elf32_arm_stub_type stub_type;
  /* Its encoding size in bytes.  */
  int stub_size;
  /* Its template.  */
  const insn_sequence *stub_template;
  /* The size of the template (number of entries).  */
  int stub_template_size;
 
  /* The symbol table entry, if any, that this was derived from.  */
  struct elf32_arm_link_hash_entry *h;
 
  /* Type of branch.  */
  enum arm_st_branch_type branch_type;
 
  /* Where this stub is being called from, or, in the case of combined
     stub sections, the first input section in the group.  */
  asection *id_sec;
 
  /* The name for the local symbol at the start of this stub.  The
     stub name in the hash table has to be unique; this does not, so
     it can be friendlier.  */
  char *output_name;
};
 
/* Used to build a map of a section.  This is required for mixed-endian
   code/data.  */
 
typedef struct elf32_elf_section_map
{
  bfd_vma vma;
  char type;
}
elf32_arm_section_map;
 
/* Information about a VFP11 erratum veneer, or a branch to such a veneer.  */
 
typedef enum
{
  VFP11_ERRATUM_BRANCH_TO_ARM_VENEER,
  VFP11_ERRATUM_BRANCH_TO_THUMB_VENEER,
  VFP11_ERRATUM_ARM_VENEER,
  VFP11_ERRATUM_THUMB_VENEER
}
elf32_vfp11_erratum_type;
 
typedef struct elf32_vfp11_erratum_list
{
  struct elf32_vfp11_erratum_list *next;
  bfd_vma vma;
  union
  {
    struct
    {
      struct elf32_vfp11_erratum_list *veneer;
      unsigned int vfp_insn;
    } b;
    struct
    {
      struct elf32_vfp11_erratum_list *branch;
      unsigned int id;
    } v;
  } u;
  elf32_vfp11_erratum_type type;
}
elf32_vfp11_erratum_list;
 
typedef enum
{
  DELETE_EXIDX_ENTRY,
  INSERT_EXIDX_CANTUNWIND_AT_END
}
arm_unwind_edit_type;
 
/* A (sorted) list of edits to apply to an unwind table.  */
typedef struct arm_unwind_table_edit
{
  arm_unwind_edit_type type;
  /* Note: we sometimes want to insert an unwind entry corresponding to a
     section different from the one we're currently writing out, so record the
     (text) section this edit relates to here.  */
  asection *linked_section;
  unsigned int index;
  struct arm_unwind_table_edit *next;
}
arm_unwind_table_edit;
 
typedef struct _arm_elf_section_data
{
  /* Information about mapping symbols.  */
  struct bfd_elf_section_data elf;
  unsigned int mapcount;
  unsigned int mapsize;
  elf32_arm_section_map *map;
  /* Information about CPU errata.  */
  unsigned int erratumcount;
  elf32_vfp11_erratum_list *erratumlist;
  /* Information about unwind tables.  */
  union
  {
    /* Unwind info attached to a text section.  */
    struct
    {
      asection *arm_exidx_sec;
    } text;
 
    /* Unwind info attached to an .ARM.exidx section.  */
    struct
    {
      arm_unwind_table_edit *unwind_edit_list;
      arm_unwind_table_edit *unwind_edit_tail;
    } exidx;
  } u;
}
_arm_elf_section_data;
 
#define elf32_arm_section_data(sec) \
  ((_arm_elf_section_data *) elf_section_data (sec))
 
/* A fix which might be required for Cortex-A8 Thumb-2 branch/TLB erratum.
   These fixes are subject to a relaxation procedure (in elf32_arm_size_stubs),
   so may be created multiple times: we use an array of these entries whilst
   relaxing which we can refresh easily, then create stubs for each potentially
   erratum-triggering instruction once we've settled on a solution.  */
 
struct a8_erratum_fix {
  bfd *input_bfd;
  asection *section;
  bfd_vma offset;
  bfd_vma addend;
  unsigned long orig_insn;
  char *stub_name;
  enum elf32_arm_stub_type stub_type;
  enum arm_st_branch_type branch_type;
};
 
/* A table of relocs applied to branches which might trigger Cortex-A8
   erratum.  */
 
struct a8_erratum_reloc {
  bfd_vma from;
  bfd_vma destination;
  struct elf32_arm_link_hash_entry *hash;
  const char *sym_name;
  unsigned int r_type;
  enum arm_st_branch_type branch_type;
  bfd_boolean non_a8_stub;
};
 
/* The size of the thread control block.  */
#define TCB_SIZE	8
 
/* ARM-specific information about a PLT entry, over and above the usual
   gotplt_union.  */
struct arm_plt_info {
  /* We reference count Thumb references to a PLT entry separately,
     so that we can emit the Thumb trampoline only if needed.  */
  bfd_signed_vma thumb_refcount;
 
  /* Some references from Thumb code may be eliminated by BL->BLX
     conversion, so record them separately.  */
  bfd_signed_vma maybe_thumb_refcount;
 
  /* How many of the recorded PLT accesses were from non-call relocations.
     This information is useful when deciding whether anything takes the
     address of an STT_GNU_IFUNC PLT.  A value of 0 means that all
     non-call references to the function should resolve directly to the
     real runtime target.  */
  unsigned int noncall_refcount;
 
  /* Since PLT entries have variable size if the Thumb prologue is
     used, we need to record the index into .got.plt instead of
     recomputing it from the PLT offset.  */
  bfd_signed_vma got_offset;
};
 
/* Information about an .iplt entry for a local STT_GNU_IFUNC symbol.  */
struct arm_local_iplt_info {
  /* The information that is usually found in the generic ELF part of
     the hash table entry.  */
  union gotplt_union root;
 
  /* The information that is usually found in the ARM-specific part of
     the hash table entry.  */
  struct arm_plt_info arm;
 
  /* A list of all potential dynamic relocations against this symbol.  */
  struct elf_dyn_relocs *dyn_relocs;
};
 
struct elf_arm_obj_tdata
{
  struct elf_obj_tdata root;
 
  /* tls_type for each local got entry.  */
  char *local_got_tls_type;
 
  /* GOTPLT entries for TLS descriptors.  */
  bfd_vma *local_tlsdesc_gotent;
 
  /* Information for local symbols that need entries in .iplt.  */
  struct arm_local_iplt_info **local_iplt;
 
  /* Zero to warn when linking objects with incompatible enum sizes.  */
  int no_enum_size_warning;
 
  /* Zero to warn when linking objects with incompatible wchar_t sizes.  */
  int no_wchar_size_warning;
};
 
#define elf_arm_tdata(bfd) \
  ((struct elf_arm_obj_tdata *) (bfd)->tdata.any)
 
#define elf32_arm_local_got_tls_type(bfd) \
  (elf_arm_tdata (bfd)->local_got_tls_type)
 
#define elf32_arm_local_tlsdesc_gotent(bfd) \
  (elf_arm_tdata (bfd)->local_tlsdesc_gotent)
 
#define elf32_arm_local_iplt(bfd) \
  (elf_arm_tdata (bfd)->local_iplt)
 
#define is_arm_elf(bfd) \
  (bfd_get_flavour (bfd) == bfd_target_elf_flavour \
   && elf_tdata (bfd) != NULL \
   && elf_object_id (bfd) == ARM_ELF_DATA)
 
static bfd_boolean
elf32_arm_mkobject (bfd *abfd)
{
  return bfd_elf_allocate_object (abfd, sizeof (struct elf_arm_obj_tdata),
				  ARM_ELF_DATA);
}
 
#define elf32_arm_hash_entry(ent) ((struct elf32_arm_link_hash_entry *)(ent))
 
/* Arm ELF linker hash entry.  */
struct elf32_arm_link_hash_entry
  {
    struct elf_link_hash_entry root;
 
    /* Track dynamic relocs copied for this symbol.  */
    struct elf_dyn_relocs *dyn_relocs;
 
    /* ARM-specific PLT information.  */
    struct arm_plt_info plt;
 
#define GOT_UNKNOWN	0
#define GOT_NORMAL	1
#define GOT_TLS_GD	2
#define GOT_TLS_IE	4
#define GOT_TLS_GDESC	8
#define GOT_TLS_GD_ANY_P(type)	((type & GOT_TLS_GD) || (type & GOT_TLS_GDESC))
    unsigned int tls_type : 8;
 
    /* True if the symbol's PLT entry is in .iplt rather than .plt.  */
    unsigned int is_iplt : 1;
 
    unsigned int unused : 23;
 
    /* Offset of the GOTPLT entry reserved for the TLS descriptor,
       starting at the end of the jump table.  */
    bfd_vma tlsdesc_got;
 
    /* The symbol marking the real symbol location for exported thumb
       symbols with Arm stubs.  */
    struct elf_link_hash_entry *export_glue;
 
   /* A pointer to the most recently used stub hash entry against this
     symbol.  */
    struct elf32_arm_stub_hash_entry *stub_cache;
  };
 
/* Traverse an arm ELF linker hash table.  */
#define elf32_arm_link_hash_traverse(table, func, info)			\
  (elf_link_hash_traverse						\
   (&(table)->root,							\
    (bfd_boolean (*) (struct elf_link_hash_entry *, void *)) (func),	\
    (info)))
 
/* Get the ARM elf linker hash table from a link_info structure.  */
#define elf32_arm_hash_table(info) \
  (elf_hash_table_id ((struct elf_link_hash_table *) ((info)->hash)) \
  == ARM_ELF_DATA ? ((struct elf32_arm_link_hash_table *) ((info)->hash)) : NULL)
 
#define arm_stub_hash_lookup(table, string, create, copy) \
  ((struct elf32_arm_stub_hash_entry *) \
   bfd_hash_lookup ((table), (string), (create), (copy)))
 
/* Array to keep track of which stub sections have been created, and
   information on stub grouping.  */
struct map_stub
{
  /* This is the section to which stubs in the group will be
     attached.  */
  asection *link_sec;
  /* The stub section.  */
  asection *stub_sec;
};
 
#define elf32_arm_compute_jump_table_size(htab) \
  ((htab)->next_tls_desc_index * 4)
 
/* ARM ELF linker hash table.  */
struct elf32_arm_link_hash_table
{
  /* The main hash table.  */
  struct elf_link_hash_table root;
 
  /* The size in bytes of the section containing the Thumb-to-ARM glue.  */
  bfd_size_type thumb_glue_size;
 
  /* The size in bytes of the section containing the ARM-to-Thumb glue.  */
  bfd_size_type arm_glue_size;
 
  /* The size in bytes of section containing the ARMv4 BX veneers.  */
  bfd_size_type bx_glue_size;
 
  /* Offsets of ARMv4 BX veneers.  Bit1 set if present, and Bit0 set when
     veneer has been populated.  */
  bfd_vma bx_glue_offset[15];
 
  /* The size in bytes of the section containing glue for VFP11 erratum
     veneers.  */
  bfd_size_type vfp11_erratum_glue_size;
 
  /* A table of fix locations for Cortex-A8 Thumb-2 branch/TLB erratum.  This
     holds Cortex-A8 erratum fix locations between elf32_arm_size_stubs() and
     elf32_arm_write_section().  */
  struct a8_erratum_fix *a8_erratum_fixes;
  unsigned int num_a8_erratum_fixes;
 
  /* An arbitrary input BFD chosen to hold the glue sections.  */
  bfd * bfd_of_glue_owner;
 
  /* Nonzero to output a BE8 image.  */
  int byteswap_code;
 
  /* Zero if R_ARM_TARGET1 means R_ARM_ABS32.
     Nonzero if R_ARM_TARGET1 means R_ARM_REL32.  */
  int target1_is_rel;
 
  /* The relocation to use for R_ARM_TARGET2 relocations.  */
  int target2_reloc;
 
  /* 0 = Ignore R_ARM_V4BX.
     1 = Convert BX to MOV PC.
     2 = Generate v4 interworing stubs.  */
  int fix_v4bx;
 
  /* Whether we should fix the Cortex-A8 Thumb-2 branch/TLB erratum.  */
  int fix_cortex_a8;
 
  /* Nonzero if the ARM/Thumb BLX instructions are available for use.  */
  int use_blx;
 
  /* What sort of code sequences we should look for which may trigger the
     VFP11 denorm erratum.  */
  bfd_arm_vfp11_fix vfp11_fix;
 
  /* Global counter for the number of fixes we have emitted.  */
  int num_vfp11_fixes;
 
  /* Nonzero to force PIC branch veneers.  */
  int pic_veneer;
 
  /* The number of bytes in the initial entry in the PLT.  */
  bfd_size_type plt_header_size;
 
  /* The number of bytes in the subsequent PLT etries.  */
  bfd_size_type plt_entry_size;
 
  /* True if the target system is VxWorks.  */
  int vxworks_p;
 
  /* True if the target system is Symbian OS.  */
  int symbian_p;
 
  /* True if the target uses REL relocations.  */
  int use_rel;
 
  /* The index of the next unused R_ARM_TLS_DESC slot in .rel.plt.  */
  bfd_vma next_tls_desc_index;
 
  /* How many R_ARM_TLS_DESC relocations were generated so far.  */
  bfd_vma num_tls_desc;
 
  /* Short-cuts to get to dynamic linker sections.  */
  asection *sdynbss;
  asection *srelbss;
 
  /* The (unloaded but important) VxWorks .rela.plt.unloaded section.  */
  asection *srelplt2;
 
  /* The offset into splt of the PLT entry for the TLS descriptor
     resolver.  Special values are 0, if not necessary (or not found
     to be necessary yet), and -1 if needed but not determined
     yet.  */
  bfd_vma dt_tlsdesc_plt;
 
  /* The offset into sgot of the GOT entry used by the PLT entry
     above.  */
  bfd_vma dt_tlsdesc_got;    
 
  /* Offset in .plt section of tls_arm_trampoline.  */
  bfd_vma tls_trampoline;
 
  /* Data for R_ARM_TLS_LDM32 relocations.  */
  union
  {
    bfd_signed_vma refcount;
    bfd_vma offset;
  } tls_ldm_got;
 
  /* Small local sym cache.  */
  struct sym_cache sym_cache;
 
  /* For convenience in allocate_dynrelocs.  */
  bfd * obfd;
 
  /* The amount of space used by the reserved portion of the sgotplt
     section, plus whatever space is used by the jump slots.  */
  bfd_vma sgotplt_jump_table_size;
 
  /* The stub hash table.  */
  struct bfd_hash_table stub_hash_table;
 
  /* Linker stub bfd.  */
  bfd *stub_bfd;
 
  /* Linker call-backs.  */
  asection * (*add_stub_section) (const char *, asection *);
  void (*layout_sections_again) (void);
 
  /* Array to keep track of which stub sections have been created, and
     information on stub grouping.  */
  struct map_stub *stub_group;
 
  /* Number of elements in stub_group.  */
  int top_id;
 
  /* Assorted information used by elf32_arm_size_stubs.  */
  unsigned int bfd_count;
  int top_index;
  asection **input_list;
};
 
/* Create an entry in an ARM ELF linker hash table.  */
 
static struct bfd_hash_entry *
elf32_arm_link_hash_newfunc (struct bfd_hash_entry * entry,
                             struct bfd_hash_table * table,
                             const char * string)
{
  struct elf32_arm_link_hash_entry * ret =
    (struct elf32_arm_link_hash_entry *) entry;
 
  /* Allocate the structure if it has not already been allocated by a
     subclass.  */
  if (ret == NULL)
    ret = (struct elf32_arm_link_hash_entry *)
        bfd_hash_allocate (table, sizeof (struct elf32_arm_link_hash_entry));
  if (ret == NULL)
    return (struct bfd_hash_entry *) ret;
 
  /* Call the allocation method of the superclass.  */
  ret = ((struct elf32_arm_link_hash_entry *)
	 _bfd_elf_link_hash_newfunc ((struct bfd_hash_entry *) ret,
				     table, string));
  if (ret != NULL)
    {
      ret->dyn_relocs = NULL;
      ret->tls_type = GOT_UNKNOWN;
      ret->tlsdesc_got = (bfd_vma) -1;
      ret->plt.thumb_refcount = 0;
      ret->plt.maybe_thumb_refcount = 0;
      ret->plt.noncall_refcount = 0;
      ret->plt.got_offset = -1;
      ret->is_iplt = FALSE;
      ret->export_glue = NULL;
 
      ret->stub_cache = NULL;
    }
 
  return (struct bfd_hash_entry *) ret;
}
 
/* Ensure that we have allocated bookkeeping structures for ABFD's local
   symbols.  */
 
static bfd_boolean
elf32_arm_allocate_local_sym_info (bfd *abfd)
{
  if (elf_local_got_refcounts (abfd) == NULL)
    {
      bfd_size_type num_syms;
      bfd_size_type size;
      char *data;
 
      num_syms = elf_tdata (abfd)->symtab_hdr.sh_info;
      size = num_syms * (sizeof (bfd_signed_vma)
			 + sizeof (struct arm_local_iplt_info *)
			 + sizeof (bfd_vma)
			 + sizeof (char));
      data = bfd_zalloc (abfd, size);
      if (data == NULL)
	return FALSE;
 
      elf_local_got_refcounts (abfd) = (bfd_signed_vma *) data;
      data += num_syms * sizeof (bfd_signed_vma);
 
      elf32_arm_local_iplt (abfd) = (struct arm_local_iplt_info **) data;
      data += num_syms * sizeof (struct arm_local_iplt_info *);
 
      elf32_arm_local_tlsdesc_gotent (abfd) = (bfd_vma *) data;
      data += num_syms * sizeof (bfd_vma);
 
      elf32_arm_local_got_tls_type (abfd) = data;
    }
  return TRUE;
}
 
/* Return the .iplt information for local symbol R_SYMNDX, which belongs
   to input bfd ABFD.  Create the information if it doesn't already exist.
   Return null if an allocation fails.  */
 
static struct arm_local_iplt_info *
elf32_arm_create_local_iplt (bfd *abfd, unsigned long r_symndx)
{
  struct arm_local_iplt_info **ptr;
 
  if (!elf32_arm_allocate_local_sym_info (abfd))
    return NULL;
 
  BFD_ASSERT (r_symndx < elf_tdata (abfd)->symtab_hdr.sh_info);
  ptr = &elf32_arm_local_iplt (abfd)[r_symndx];
  if (*ptr == NULL)
    *ptr = bfd_zalloc (abfd, sizeof (**ptr));
  return *ptr;
}
 
/* Try to obtain PLT information for the symbol with index R_SYMNDX
   in ABFD's symbol table.  If the symbol is global, H points to its
   hash table entry, otherwise H is null.
 
   Return true if the symbol does have PLT information.  When returning
   true, point *ROOT_PLT at the target-independent reference count/offset
   union and *ARM_PLT at the ARM-specific information.  */
 
static bfd_boolean
elf32_arm_get_plt_info (bfd *abfd, struct elf32_arm_link_hash_entry *h,
			unsigned long r_symndx, union gotplt_union **root_plt,
			struct arm_plt_info **arm_plt)
{
  struct arm_local_iplt_info *local_iplt;
 
  if (h != NULL)
    {
      *root_plt = &h->root.plt;
      *arm_plt = &h->plt;
      return TRUE;
    }
 
  if (elf32_arm_local_iplt (abfd) == NULL)
    return FALSE;
 
  local_iplt = elf32_arm_local_iplt (abfd)[r_symndx];
  if (local_iplt == NULL)
    return FALSE;
 
  *root_plt = &local_iplt->root;
  *arm_plt = &local_iplt->arm;
  return TRUE;
}
 
/* Return true if the PLT described by ARM_PLT requires a Thumb stub
   before it.  */
 
static bfd_boolean
elf32_arm_plt_needs_thumb_stub_p (struct bfd_link_info *info,
				  struct arm_plt_info *arm_plt)
{
  struct elf32_arm_link_hash_table *htab;
 
  htab = elf32_arm_hash_table (info);
  return (arm_plt->thumb_refcount != 0
	  || (!htab->use_blx && arm_plt->maybe_thumb_refcount != 0));
}
 
/* Return a pointer to the head of the dynamic reloc list that should
   be used for local symbol ISYM, which is symbol number R_SYMNDX in
   ABFD's symbol table.  Return null if an error occurs.  */
 
static struct elf_dyn_relocs **
elf32_arm_get_local_dynreloc_list (bfd *abfd, unsigned long r_symndx,
				   Elf_Internal_Sym *isym)
{
  if (ELF32_ST_TYPE (isym->st_info) == STT_GNU_IFUNC)
    {
      struct arm_local_iplt_info *local_iplt;
 
      local_iplt = elf32_arm_create_local_iplt (abfd, r_symndx);
      if (local_iplt == NULL)
	return NULL;
      return &local_iplt->dyn_relocs;
    }
  else
    {
      /* Track dynamic relocs needed for local syms too.
	 We really need local syms available to do this
	 easily.  Oh well.  */
      asection *s;
      void *vpp;
 
      s = bfd_section_from_elf_index (abfd, isym->st_shndx);
      if (s == NULL)
	abort ();
 
      vpp = &elf_section_data (s)->local_dynrel;
      return (struct elf_dyn_relocs **) vpp;
    }
}
 
/* Initialize an entry in the stub hash table.  */
 
static struct bfd_hash_entry *
stub_hash_newfunc (struct bfd_hash_entry *entry,
		   struct bfd_hash_table *table,
		   const char *string)
{
  /* Allocate the structure if it has not already been allocated by a
     subclass.  */
  if (entry == NULL)
    {
      entry = (struct bfd_hash_entry *)
          bfd_hash_allocate (table, sizeof (struct elf32_arm_stub_hash_entry));
      if (entry == NULL)
	return entry;
    }
 
  /* Call the allocation method of the superclass.  */
  entry = bfd_hash_newfunc (entry, table, string);
  if (entry != NULL)
    {
      struct elf32_arm_stub_hash_entry *eh;
 
      /* Initialize the local fields.  */
      eh = (struct elf32_arm_stub_hash_entry *) entry;
      eh->stub_sec = NULL;
      eh->stub_offset = 0;
      eh->target_value = 0;
      eh->target_section = NULL;
      eh->target_addend = 0;
      eh->orig_insn = 0;
      eh->stub_type = arm_stub_none;
      eh->stub_size = 0;
      eh->stub_template = NULL;
      eh->stub_template_size = 0;
      eh->h = NULL;
      eh->id_sec = NULL;
      eh->output_name = NULL;
    }
 
  return entry;
}
 
/* Create .got, .gotplt, and .rel(a).got sections in DYNOBJ, and set up
   shortcuts to them in our hash table.  */
 
static bfd_boolean
create_got_section (bfd *dynobj, struct bfd_link_info *info)
{
  struct elf32_arm_link_hash_table *htab;
 
  htab = elf32_arm_hash_table (info);
  if (htab == NULL)
    return FALSE;
 
  /* BPABI objects never have a GOT, or associated sections.  */
  if (htab->symbian_p)
    return TRUE;
 
  if (! _bfd_elf_create_got_section (dynobj, info))
    return FALSE;
 
  return TRUE;
}
 
/* Create the .iplt, .rel(a).iplt and .igot.plt sections.  */
 
static bfd_boolean
create_ifunc_sections (struct bfd_link_info *info)
{
  struct elf32_arm_link_hash_table *htab;
  const struct elf_backend_data *bed;
  bfd *dynobj;
  asection *s;
  flagword flags;
 
  htab = elf32_arm_hash_table (info);
  dynobj = htab->root.dynobj;
  bed = get_elf_backend_data (dynobj);
  flags = bed->dynamic_sec_flags;
 
  if (htab->root.iplt == NULL)
    {
      s = bfd_make_section_with_flags (dynobj, ".iplt",
				       flags | SEC_READONLY | SEC_CODE);
      if (s == NULL
	  || !bfd_set_section_alignment (abfd, s, bed->plt_alignment))
	return FALSE;
      htab->root.iplt = s;
    }
 
  if (htab->root.irelplt == NULL)
    {
      s = bfd_make_section_with_flags (dynobj, RELOC_SECTION (htab, ".iplt"),
				       flags | SEC_READONLY);
      if (s == NULL
	  || !bfd_set_section_alignment (abfd, s, bed->s->log_file_align))
	return FALSE;
      htab->root.irelplt = s;
    }
 
  if (htab->root.igotplt == NULL)
    {
      s = bfd_make_section_with_flags (dynobj, ".igot.plt", flags);
      if (s == NULL
	  || !bfd_set_section_alignment (dynobj, s, bed->s->log_file_align))
	return FALSE;
      htab->root.igotplt = s;
    }
  return TRUE;
}
 
/* Create .plt, .rel(a).plt, .got, .got.plt, .rel(a).got, .dynbss, and
   .rel(a).bss sections in DYNOBJ, and set up shortcuts to them in our
   hash table.  */
 
static bfd_boolean
elf32_arm_create_dynamic_sections (bfd *dynobj, struct bfd_link_info *info)
{
  struct elf32_arm_link_hash_table *htab;
 
  htab = elf32_arm_hash_table (info);
  if (htab == NULL)
    return FALSE;
 
  if (!htab->root.sgot && !create_got_section (dynobj, info))
    return FALSE;
 
  if (!_bfd_elf_create_dynamic_sections (dynobj, info))
    return FALSE;
 
  htab->sdynbss = bfd_get_section_by_name (dynobj, ".dynbss");
  if (!info->shared)
    htab->srelbss = bfd_get_section_by_name (dynobj,
					     RELOC_SECTION (htab, ".bss"));
 
  if (htab->vxworks_p)
    {
      if (!elf_vxworks_create_dynamic_sections (dynobj, info, &htab->srelplt2))
	return FALSE;
 
      if (info->shared)
	{
	  htab->plt_header_size = 0;
	  htab->plt_entry_size
	    = 4 * ARRAY_SIZE (elf32_arm_vxworks_shared_plt_entry);
	}
      else
	{
	  htab->plt_header_size
	    = 4 * ARRAY_SIZE (elf32_arm_vxworks_exec_plt0_entry);
	  htab->plt_entry_size
	    = 4 * ARRAY_SIZE (elf32_arm_vxworks_exec_plt_entry);
	}
    }
 
  if (!htab->root.splt
      || !htab->root.srelplt
      || !htab->sdynbss
      || (!info->shared && !htab->srelbss))
    abort ();
 
  return TRUE;
}
 
/* Copy the extra info we tack onto an elf_link_hash_entry.  */
 
static void
elf32_arm_copy_indirect_symbol (struct bfd_link_info *info,
				struct elf_link_hash_entry *dir,
				struct elf_link_hash_entry *ind)
{
  struct elf32_arm_link_hash_entry *edir, *eind;
 
  edir = (struct elf32_arm_link_hash_entry *) dir;
  eind = (struct elf32_arm_link_hash_entry *) ind;
 
  if (eind->dyn_relocs != NULL)
    {
      if (edir->dyn_relocs != NULL)
	{
	  struct elf_dyn_relocs **pp;
	  struct elf_dyn_relocs *p;
 
	  /* Add reloc counts against the indirect sym to the direct sym
	     list.  Merge any entries against the same section.  */
	  for (pp = &eind->dyn_relocs; (p = *pp) != NULL; )
	    {
	      struct elf_dyn_relocs *q;
 
	      for (q = edir->dyn_relocs; q != NULL; q = q->next)
		if (q->sec == p->sec)
		  {
		    q->pc_count += p->pc_count;
		    q->count += p->count;
		    *pp = p->next;
		    break;
		  }
	      if (q == NULL)
		pp = &p->next;
	    }
	  *pp = edir->dyn_relocs;
	}
 
      edir->dyn_relocs = eind->dyn_relocs;
      eind->dyn_relocs = NULL;
    }
 
  if (ind->root.type == bfd_link_hash_indirect)
    {
      /* Copy over PLT info.  */
      edir->plt.thumb_refcount += eind->plt.thumb_refcount;
      eind->plt.thumb_refcount = 0;
      edir->plt.maybe_thumb_refcount += eind->plt.maybe_thumb_refcount;
      eind->plt.maybe_thumb_refcount = 0;
      edir->plt.noncall_refcount += eind->plt.noncall_refcount;
      eind->plt.noncall_refcount = 0;
 
      /* We should only allocate a function to .iplt once the final
	 symbol information is known.  */
      BFD_ASSERT (!eind->is_iplt);
 
      if (dir->got.refcount <= 0)
	{
	  edir->tls_type = eind->tls_type;
	  eind->tls_type = GOT_UNKNOWN;
	}
    }
 
  _bfd_elf_link_hash_copy_indirect (info, dir, ind);
}
 
/* Create an ARM elf linker hash table.  */
 
static struct bfd_link_hash_table *
elf32_arm_link_hash_table_create (bfd *abfd)
{
  struct elf32_arm_link_hash_table *ret;
  bfd_size_type amt = sizeof (struct elf32_arm_link_hash_table);
 
  ret = (struct elf32_arm_link_hash_table *) bfd_malloc (amt);
  if (ret == NULL)
    return NULL;
 
  if (!_bfd_elf_link_hash_table_init (& ret->root, abfd,
				      elf32_arm_link_hash_newfunc,
				      sizeof (struct elf32_arm_link_hash_entry),
				      ARM_ELF_DATA))
    {
      free (ret);
      return NULL;
    }
 
  ret->sdynbss = NULL;
  ret->srelbss = NULL;
  ret->srelplt2 = NULL;
  ret->dt_tlsdesc_plt = 0;
  ret->dt_tlsdesc_got = 0;
  ret->tls_trampoline = 0;
  ret->next_tls_desc_index = 0;
  ret->num_tls_desc = 0;
  ret->thumb_glue_size = 0;
  ret->arm_glue_size = 0;
  ret->bx_glue_size = 0;
  memset (ret->bx_glue_offset, 0, sizeof (ret->bx_glue_offset));
  ret->vfp11_fix = BFD_ARM_VFP11_FIX_NONE;
  ret->vfp11_erratum_glue_size = 0;
  ret->num_vfp11_fixes = 0;
  ret->fix_cortex_a8 = 0;
  ret->bfd_of_glue_owner = NULL;
  ret->byteswap_code = 0;
  ret->target1_is_rel = 0;
  ret->target2_reloc = R_ARM_NONE;
#ifdef FOUR_WORD_PLT
  ret->plt_header_size = 16;
  ret->plt_entry_size = 16;
#else
  ret->plt_header_size = 20;
  ret->plt_entry_size = 12;
#endif
  ret->fix_v4bx = 0;
  ret->use_blx = 0;
  ret->vxworks_p = 0;
  ret->symbian_p = 0;
  ret->use_rel = 1;
  ret->sym_cache.abfd = NULL;
  ret->obfd = abfd;
  ret->tls_ldm_got.refcount = 0;
  ret->stub_bfd = NULL;
  ret->add_stub_section = NULL;
  ret->layout_sections_again = NULL;
  ret->stub_group = NULL;
  ret->top_id = 0;
  ret->bfd_count = 0;
  ret->top_index = 0;
  ret->input_list = NULL;
 
  if (!bfd_hash_table_init (&ret->stub_hash_table, stub_hash_newfunc,
			    sizeof (struct elf32_arm_stub_hash_entry)))
    {
      free (ret);
      return NULL;
    }
 
  return &ret->root.root;
}
 
/* Free the derived linker hash table.  */
 
static void
elf32_arm_hash_table_free (struct bfd_link_hash_table *hash)
{
  struct elf32_arm_link_hash_table *ret
    = (struct elf32_arm_link_hash_table *) hash;
 
  bfd_hash_table_free (&ret->stub_hash_table);
  _bfd_generic_link_hash_table_free (hash);
}
 
/* Determine if we're dealing with a Thumb only architecture.  */
 
static bfd_boolean
using_thumb_only (struct elf32_arm_link_hash_table *globals)
{
  int arch = bfd_elf_get_obj_attr_int (globals->obfd, OBJ_ATTR_PROC,
				       Tag_CPU_arch);
  int profile;
 
  if (arch == TAG_CPU_ARCH_V6_M || arch == TAG_CPU_ARCH_V6S_M)
    return TRUE;
 
  if (arch != TAG_CPU_ARCH_V7 && arch != TAG_CPU_ARCH_V7E_M)
    return FALSE;
 
  profile = bfd_elf_get_obj_attr_int (globals->obfd, OBJ_ATTR_PROC,
				      Tag_CPU_arch_profile);
 
  return profile == 'M';
}
 
/* Determine if we're dealing with a Thumb-2 object.  */
 
static bfd_boolean
using_thumb2 (struct elf32_arm_link_hash_table *globals)
{
  int arch = bfd_elf_get_obj_attr_int (globals->obfd, OBJ_ATTR_PROC,
				       Tag_CPU_arch);
  return arch == TAG_CPU_ARCH_V6T2 || arch >= TAG_CPU_ARCH_V7;
}
 
/* Determine what kind of NOPs are available.  */
 
static bfd_boolean
arch_has_arm_nop (struct elf32_arm_link_hash_table *globals)
{
  const int arch = bfd_elf_get_obj_attr_int (globals->obfd, OBJ_ATTR_PROC,
					     Tag_CPU_arch);
  return arch == TAG_CPU_ARCH_V6T2
	 || arch == TAG_CPU_ARCH_V6K
	 || arch == TAG_CPU_ARCH_V7
	 || arch == TAG_CPU_ARCH_V7E_M;
}
 
static bfd_boolean
arch_has_thumb2_nop (struct elf32_arm_link_hash_table *globals)
{
  const int arch = bfd_elf_get_obj_attr_int (globals->obfd, OBJ_ATTR_PROC,
					     Tag_CPU_arch);
  return (arch == TAG_CPU_ARCH_V6T2 || arch == TAG_CPU_ARCH_V7
	  || arch == TAG_CPU_ARCH_V7E_M);
}
 
static bfd_boolean
arm_stub_is_thumb (enum elf32_arm_stub_type stub_type)
{
  switch (stub_type)
    {
    case arm_stub_long_branch_thumb_only:
    case arm_stub_long_branch_v4t_thumb_arm:
    case arm_stub_short_branch_v4t_thumb_arm:
    case arm_stub_long_branch_v4t_thumb_arm_pic:
    case arm_stub_long_branch_v4t_thumb_tls_pic:
    case arm_stub_long_branch_thumb_only_pic:
      return TRUE;
    case arm_stub_none:
      BFD_FAIL ();
      return FALSE;
      break;
    default:
      return FALSE;
    }
}
 
/* Determine the type of stub needed, if any, for a call.  */
 
static enum elf32_arm_stub_type
arm_type_of_stub (struct bfd_link_info *info,
		  asection *input_sec,
		  const Elf_Internal_Rela *rel,
		  unsigned char st_type,
		  enum arm_st_branch_type *actual_branch_type,
		  struct elf32_arm_link_hash_entry *hash,
		  bfd_vma destination,
		  asection *sym_sec,
		  bfd *input_bfd,
		  const char *name)
{
  bfd_vma location;
  bfd_signed_vma branch_offset;
  unsigned int r_type;
  struct elf32_arm_link_hash_table * globals;
  int thumb2;
  int thumb_only;
  enum elf32_arm_stub_type stub_type = arm_stub_none;
  int use_plt = 0;
  enum arm_st_branch_type branch_type = *actual_branch_type;
  union gotplt_union *root_plt;
  struct arm_plt_info *arm_plt;
 
  if (branch_type == ST_BRANCH_LONG)
    return stub_type;
 
  globals = elf32_arm_hash_table (info);
  if (globals == NULL)
    return stub_type;
 
  thumb_only = using_thumb_only (globals);
 
  thumb2 = using_thumb2 (globals);
 
  /* Determine where the call point is.  */
  location = (input_sec->output_offset
	      + input_sec->output_section->vma
	      + rel->r_offset);
 
  r_type = ELF32_R_TYPE (rel->r_info);
 
  /* For TLS call relocs, it is the caller's responsibility to provide
     the address of the appropriate trampoline.  */
  if (r_type != R_ARM_TLS_CALL
      && r_type != R_ARM_THM_TLS_CALL
      && elf32_arm_get_plt_info (input_bfd, hash, ELF32_R_SYM (rel->r_info),
				 &root_plt, &arm_plt)
      && root_plt->offset != (bfd_vma) -1)
    {
      asection *splt;
 
      if (hash == NULL || hash->is_iplt)
	splt = globals->root.iplt;
      else
	splt = globals->root.splt;
      if (splt != NULL)
	{	
	  use_plt = 1;
 
	  /* Note when dealing with PLT entries: the main PLT stub is in
	     ARM mode, so if the branch is in Thumb mode, another
	     Thumb->ARM stub will be inserted later just before the ARM
	     PLT stub. We don't take this extra distance into account
	     here, because if a long branch stub is needed, we'll add a
	     Thumb->Arm one and branch directly to the ARM PLT entry
	     because it avoids spreading offset corrections in several
	     places.  */
 
	  destination = (splt->output_section->vma
			 + splt->output_offset
			 + root_plt->offset);
	  st_type = STT_FUNC;
	  branch_type = ST_BRANCH_TO_ARM;
	}
    }
  /* Calls to STT_GNU_IFUNC symbols should go through a PLT.  */
  BFD_ASSERT (st_type != STT_GNU_IFUNC);
 
  branch_offset = (bfd_signed_vma)(destination - location);
 
  if (r_type == R_ARM_THM_CALL || r_type == R_ARM_THM_JUMP24
      || r_type == R_ARM_THM_TLS_CALL)
    {
      /* Handle cases where:
	 - this call goes too far (different Thumb/Thumb2 max
           distance)
	 - it's a Thumb->Arm call and blx is not available, or it's a
           Thumb->Arm branch (not bl). A stub is needed in this case,
           but only if this call is not through a PLT entry. Indeed,
           PLT stubs handle mode switching already.
      */
      if ((!thumb2
	    && (branch_offset > THM_MAX_FWD_BRANCH_OFFSET
		|| (branch_offset < THM_MAX_BWD_BRANCH_OFFSET)))
	  || (thumb2
	      && (branch_offset > THM2_MAX_FWD_BRANCH_OFFSET
		  || (branch_offset < THM2_MAX_BWD_BRANCH_OFFSET)))
	  || (branch_type == ST_BRANCH_TO_ARM
	      && (((r_type == R_ARM_THM_CALL
		    || r_type == R_ARM_THM_TLS_CALL) && !globals->use_blx)
		  || (r_type == R_ARM_THM_JUMP24))
	      && !use_plt))
	{
	  if (branch_type == ST_BRANCH_TO_THUMB)
	    {
	      /* Thumb to thumb.  */
	      if (!thumb_only)
		{
		  stub_type = (info->shared | globals->pic_veneer)
		    /* PIC stubs.  */
		    ? ((globals->use_blx
			&& (r_type ==R_ARM_THM_CALL))
		       /* V5T and above. Stub starts with ARM code, so
			  we must be able to switch mode before
			  reaching it, which is only possible for 'bl'
			  (ie R_ARM_THM_CALL relocation).  */
		       ? arm_stub_long_branch_any_thumb_pic
		       /* On V4T, use Thumb code only.  */
		       : arm_stub_long_branch_v4t_thumb_thumb_pic)
 
		    /* non-PIC stubs.  */
		    : ((globals->use_blx
			&& (r_type ==R_ARM_THM_CALL))
		       /* V5T and above.  */
		       ? arm_stub_long_branch_any_any
		       /* V4T.  */
		       : arm_stub_long_branch_v4t_thumb_thumb);
		}
	      else
		{
		  stub_type = (info->shared | globals->pic_veneer)
		    /* PIC stub.  */
		    ? arm_stub_long_branch_thumb_only_pic
		    /* non-PIC stub.  */
		    : arm_stub_long_branch_thumb_only;
		}
	    }
	  else
	    {
	      /* Thumb to arm.  */
	      if (sym_sec != NULL
		  && sym_sec->owner != NULL
		  && !INTERWORK_FLAG (sym_sec->owner))
		{
		  (*_bfd_error_handler)
		    (_("%B(%s): warning: interworking not enabled.\n"
		       "  first occurrence: %B: Thumb call to ARM"),
		     sym_sec->owner, input_bfd, name);
		}
 
	      stub_type =
		(info->shared | globals->pic_veneer)
		/* PIC stubs.  */
		? (r_type == R_ARM_THM_TLS_CALL
		   /* TLS PIC stubs */
		   ? (globals->use_blx ? arm_stub_long_branch_any_tls_pic
		      : arm_stub_long_branch_v4t_thumb_tls_pic)
		   : ((globals->use_blx && r_type == R_ARM_THM_CALL)
		      /* V5T PIC and above.  */
		      ? arm_stub_long_branch_any_arm_pic
		      /* V4T PIC stub.  */
		      : arm_stub_long_branch_v4t_thumb_arm_pic))
 
		/* non-PIC stubs.  */
		: ((globals->use_blx && r_type == R_ARM_THM_CALL)
		   /* V5T and above.  */
		   ? arm_stub_long_branch_any_any
		   /* V4T.  */
		   : arm_stub_long_branch_v4t_thumb_arm);
 
	      /* Handle v4t short branches.  */
	      if ((stub_type == arm_stub_long_branch_v4t_thumb_arm)
		  && (branch_offset <= THM_MAX_FWD_BRANCH_OFFSET)
		  && (branch_offset >= THM_MAX_BWD_BRANCH_OFFSET))
		stub_type = arm_stub_short_branch_v4t_thumb_arm;
	    }
	}
    }
  else if (r_type == R_ARM_CALL
	   || r_type == R_ARM_JUMP24
	   || r_type == R_ARM_PLT32
	   || r_type == R_ARM_TLS_CALL)
    {
      if (branch_type == ST_BRANCH_TO_THUMB)
	{
	  /* Arm to thumb.  */
 
	  if (sym_sec != NULL
	      && sym_sec->owner != NULL
	      && !INTERWORK_FLAG (sym_sec->owner))
	    {
	      (*_bfd_error_handler)
		(_("%B(%s): warning: interworking not enabled.\n"
		   "  first occurrence: %B: ARM call to Thumb"),
		 sym_sec->owner, input_bfd, name);
	    }
 
	  /* We have an extra 2-bytes reach because of
	     the mode change (bit 24 (H) of BLX encoding).  */
	  if (branch_offset > (ARM_MAX_FWD_BRANCH_OFFSET + 2)
	      || (branch_offset < ARM_MAX_BWD_BRANCH_OFFSET)
	      || (r_type == R_ARM_CALL && !globals->use_blx)
	      || (r_type == R_ARM_JUMP24)
	      || (r_type == R_ARM_PLT32))
	    {
	      stub_type = (info->shared | globals->pic_veneer)
		/* PIC stubs.  */
		? ((globals->use_blx)
		   /* V5T and above.  */
		   ? arm_stub_long_branch_any_thumb_pic
		   /* V4T stub.  */
		   : arm_stub_long_branch_v4t_arm_thumb_pic)
 
		/* non-PIC stubs.  */
		: ((globals->use_blx)
		   /* V5T and above.  */
		   ? arm_stub_long_branch_any_any
		   /* V4T.  */
		   : arm_stub_long_branch_v4t_arm_thumb);
	    }
	}
      else
	{
	  /* Arm to arm.  */
	  if (branch_offset > ARM_MAX_FWD_BRANCH_OFFSET
	      || (branch_offset < ARM_MAX_BWD_BRANCH_OFFSET))
	    {
	      stub_type =
		(info->shared | globals->pic_veneer)
		/* PIC stubs.  */
		? (r_type == R_ARM_TLS_CALL
		   /* TLS PIC Stub */
		   ? arm_stub_long_branch_any_tls_pic
		   : arm_stub_long_branch_any_arm_pic)
		/* non-PIC stubs.  */
		: arm_stub_long_branch_any_any;
	    }
	}
    }
 
  /* If a stub is needed, record the actual destination type.  */
  if (stub_type != arm_stub_none)
    *actual_branch_type = branch_type;
 
  return stub_type;
}
 
/* Build a name for an entry in the stub hash table.  */
 
static char *
elf32_arm_stub_name (const asection *input_section,
		     const asection *sym_sec,
		     const struct elf32_arm_link_hash_entry *hash,
		     const Elf_Internal_Rela *rel,
		     enum elf32_arm_stub_type stub_type)
{
  char *stub_name;
  bfd_size_type len;
 
  if (hash)
    {
      len = 8 + 1 + strlen (hash->root.root.root.string) + 1 + 8 + 1 + 2 + 1;
      stub_name = (char *) bfd_malloc (len);
      if (stub_name != NULL)
	sprintf (stub_name, "%08x_%s+%x_%d",
		 input_section->id & 0xffffffff,
		 hash->root.root.root.string,
		 (int) rel->r_addend & 0xffffffff,
		 (int) stub_type);
    }
  else
    {
      len = 8 + 1 + 8 + 1 + 8 + 1 + 8 + 1 + 2 + 1;
      stub_name = (char *) bfd_malloc (len);
      if (stub_name != NULL)
	sprintf (stub_name, "%08x_%x:%x+%x_%d",
		 input_section->id & 0xffffffff,
		 sym_sec->id & 0xffffffff,
		 ELF32_R_TYPE (rel->r_info) == R_ARM_TLS_CALL
		 || ELF32_R_TYPE (rel->r_info) == R_ARM_THM_TLS_CALL
		 ? 0 : (int) ELF32_R_SYM (rel->r_info) & 0xffffffff,
		 (int) rel->r_addend & 0xffffffff,
		 (int) stub_type);
    }
 
  return stub_name;
}
 
/* Look up an entry in the stub hash.  Stub entries are cached because
   creating the stub name takes a bit of time.  */
 
static struct elf32_arm_stub_hash_entry *
elf32_arm_get_stub_entry (const asection *input_section,
			  const asection *sym_sec,
			  struct elf_link_hash_entry *hash,
			  const Elf_Internal_Rela *rel,
			  struct elf32_arm_link_hash_table *htab,
			  enum elf32_arm_stub_type stub_type)
{
  struct elf32_arm_stub_hash_entry *stub_entry;
  struct elf32_arm_link_hash_entry *h = (struct elf32_arm_link_hash_entry *) hash;
  const asection *id_sec;
 
  if ((input_section->flags & SEC_CODE) == 0)
    return NULL;
 
  /* If this input section is part of a group of sections sharing one
     stub section, then use the id of the first section in the group.
     Stub names need to include a section id, as there may well be
     more than one stub used to reach say, printf, and we need to
     distinguish between them.  */
  id_sec = htab->stub_group[input_section->id].link_sec;
 
  if (h != NULL && h->stub_cache != NULL
      && h->stub_cache->h == h
      && h->stub_cache->id_sec == id_sec
      && h->stub_cache->stub_type == stub_type)
    {
      stub_entry = h->stub_cache;
    }
  else
    {
      char *stub_name;
 
      stub_name = elf32_arm_stub_name (id_sec, sym_sec, h, rel, stub_type);
      if (stub_name == NULL)
	return NULL;
 
      stub_entry = arm_stub_hash_lookup (&htab->stub_hash_table,
					stub_name, FALSE, FALSE);
      if (h != NULL)
	h->stub_cache = stub_entry;
 
      free (stub_name);
    }
 
  return stub_entry;
}
 
/* Find or create a stub section.  Returns a pointer to the stub section, and
   the section to which the stub section will be attached (in *LINK_SEC_P). 
   LINK_SEC_P may be NULL.  */
 
static asection *
elf32_arm_create_or_find_stub_sec (asection **link_sec_p, asection *section,
				   struct elf32_arm_link_hash_table *htab)
{
  asection *link_sec;
  asection *stub_sec;
 
  link_sec = htab->stub_group[section->id].link_sec;
  stub_sec = htab->stub_group[section->id].stub_sec;
  if (stub_sec == NULL)
    {
      stub_sec = htab->stub_group[link_sec->id].stub_sec;
      if (stub_sec == NULL)
	{
	  size_t namelen;
	  bfd_size_type len;
	  char *s_name;
 
	  namelen = strlen (link_sec->name);
	  len = namelen + sizeof (STUB_SUFFIX);
	  s_name = (char *) bfd_alloc (htab->stub_bfd, len);
	  if (s_name == NULL)
	    return NULL;
 
	  memcpy (s_name, link_sec->name, namelen);
	  memcpy (s_name + namelen, STUB_SUFFIX, sizeof (STUB_SUFFIX));
	  stub_sec = (*htab->add_stub_section) (s_name, link_sec);
	  if (stub_sec == NULL)
	    return NULL;
	  htab->stub_group[link_sec->id].stub_sec = stub_sec;
	}
      htab->stub_group[section->id].stub_sec = stub_sec;
    }
 
  if (link_sec_p)
    *link_sec_p = link_sec;
 
  return stub_sec;
}
 
/* Add a new stub entry to the stub hash.  Not all fields of the new
   stub entry are initialised.  */
 
static struct elf32_arm_stub_hash_entry *
elf32_arm_add_stub (const char *stub_name,
		    asection *section,
		    struct elf32_arm_link_hash_table *htab)
{
  asection *link_sec;
  asection *stub_sec;
  struct elf32_arm_stub_hash_entry *stub_entry;
 
  stub_sec = elf32_arm_create_or_find_stub_sec (&link_sec, section, htab);
  if (stub_sec == NULL)
    return NULL;
 
  /* Enter this entry into the linker stub hash table.  */
  stub_entry = arm_stub_hash_lookup (&htab->stub_hash_table, stub_name,
				     TRUE, FALSE);
  if (stub_entry == NULL)
    {
      (*_bfd_error_handler) (_("%s: cannot create stub entry %s"),
			     section->owner,
			     stub_name);
      return NULL;
    }
 
  stub_entry->stub_sec = stub_sec;
  stub_entry->stub_offset = 0;
  stub_entry->id_sec = link_sec;
 
  return stub_entry;
}
 
/* Store an Arm insn into an output section not processed by
   elf32_arm_write_section.  */
 
static void
put_arm_insn (struct elf32_arm_link_hash_table * htab,
	      bfd * output_bfd, bfd_vma val, void * ptr)
{
  if (htab->byteswap_code != bfd_little_endian (output_bfd))
    bfd_putl32 (val, ptr);
  else
    bfd_putb32 (val, ptr);
}
 
/* Store a 16-bit Thumb insn into an output section not processed by
   elf32_arm_write_section.  */
 
static void
put_thumb_insn (struct elf32_arm_link_hash_table * htab,
		bfd * output_bfd, bfd_vma val, void * ptr)
{
  if (htab->byteswap_code != bfd_little_endian (output_bfd))
    bfd_putl16 (val, ptr);
  else
    bfd_putb16 (val, ptr);
}
 
/* If it's possible to change R_TYPE to a more efficient access
   model, return the new reloc type.  */
 
static unsigned
elf32_arm_tls_transition (struct bfd_link_info *info, int r_type, 
			  struct elf_link_hash_entry *h)
{
  int is_local = (h == NULL);
 
  if (info->shared || (h && h->root.type == bfd_link_hash_undefweak))
    return r_type;
 
  /* We do not support relaxations for Old TLS models.  */ 
  switch (r_type)
    {
    case R_ARM_TLS_GOTDESC:
    case R_ARM_TLS_CALL:
    case R_ARM_THM_TLS_CALL:
    case R_ARM_TLS_DESCSEQ:
    case R_ARM_THM_TLS_DESCSEQ:
      return is_local ? R_ARM_TLS_LE32 : R_ARM_TLS_IE32;
    }
 
  return r_type;
}
 
static bfd_reloc_status_type elf32_arm_final_link_relocate
  (reloc_howto_type *, bfd *, bfd *, asection *, bfd_byte *,
   Elf_Internal_Rela *, bfd_vma, struct bfd_link_info *, asection *,
   const char *, unsigned char, enum arm_st_branch_type,
   struct elf_link_hash_entry *, bfd_boolean *, char **);
 
static unsigned int
arm_stub_required_alignment (enum elf32_arm_stub_type stub_type)
{
  switch (stub_type)
    {
    case arm_stub_a8_veneer_b_cond:
    case arm_stub_a8_veneer_b:
    case arm_stub_a8_veneer_bl:
      return 2;
 
    case arm_stub_long_branch_any_any:
    case arm_stub_long_branch_v4t_arm_thumb:
    case arm_stub_long_branch_thumb_only:
    case arm_stub_long_branch_v4t_thumb_thumb:
    case arm_stub_long_branch_v4t_thumb_arm:
    case arm_stub_short_branch_v4t_thumb_arm:
    case arm_stub_long_branch_any_arm_pic:
    case arm_stub_long_branch_any_thumb_pic:
    case arm_stub_long_branch_v4t_thumb_thumb_pic:
    case arm_stub_long_branch_v4t_arm_thumb_pic:
    case arm_stub_long_branch_v4t_thumb_arm_pic:
    case arm_stub_long_branch_thumb_only_pic:
    case arm_stub_long_branch_any_tls_pic:
    case arm_stub_long_branch_v4t_thumb_tls_pic:
    case arm_stub_a8_veneer_blx:
      return 4;
 
    default:
      abort ();  /* Should be unreachable.  */
    }
}
 
static bfd_boolean
arm_build_one_stub (struct bfd_hash_entry *gen_entry,
		    void * in_arg)
{
#define MAXRELOCS 2
  struct elf32_arm_stub_hash_entry *stub_entry;
  struct elf32_arm_link_hash_table *globals;
  struct bfd_link_info *info;
  asection *stub_sec;
  bfd *stub_bfd;
  bfd_byte *loc;
  bfd_vma sym_value;
  int template_size;
  int size;
  const insn_sequence *template_sequence;
  int i;
  int stub_reloc_idx[MAXRELOCS] = {-1, -1};
  int stub_reloc_offset[MAXRELOCS] = {0, 0};
  int nrelocs = 0;
 
  /* Massage our args to the form they really have.  */
  stub_entry = (struct elf32_arm_stub_hash_entry *) gen_entry;
  info = (struct bfd_link_info *) in_arg;
 
  globals = elf32_arm_hash_table (info);
  if (globals == NULL)
    return FALSE;
 
  stub_sec = stub_entry->stub_sec;
 
  if ((globals->fix_cortex_a8 < 0)
      != (arm_stub_required_alignment (stub_entry->stub_type) == 2))
    /* We have to do less-strictly-aligned fixes last.  */
    return TRUE;
 
  /* Make a note of the offset within the stubs for this entry.  */
  stub_entry->stub_offset = stub_sec->size;
  loc = stub_sec->contents + stub_entry->stub_offset;
 
  stub_bfd = stub_sec->owner;
 
  /* This is the address of the stub destination.  */
  sym_value = (stub_entry->target_value
	       + stub_entry->target_section->output_offset
	       + stub_entry->target_section->output_section->vma);
 
  template_sequence = stub_entry->stub_template;
  template_size = stub_entry->stub_template_size;
 
  size = 0;
  for (i = 0; i < template_size; i++)
    {
      switch (template_sequence[i].type)
	{
	case THUMB16_TYPE:
	  {
	    bfd_vma data = (bfd_vma) template_sequence[i].data;
	    if (template_sequence[i].reloc_addend != 0)
	      {
                /* We've borrowed the reloc_addend field to mean we should
                   insert a condition code into this (Thumb-1 branch)
                   instruction.  See THUMB16_BCOND_INSN.  */
                BFD_ASSERT ((data & 0xff00) == 0xd000);
                data |= ((stub_entry->orig_insn >> 22) & 0xf) << 8;
	      }
	    bfd_put_16 (stub_bfd, data, loc + size);
	    size += 2;
	  }
	  break;
 
	case THUMB32_TYPE:
	  bfd_put_16 (stub_bfd,
		      (template_sequence[i].data >> 16) & 0xffff,
		      loc + size);
	  bfd_put_16 (stub_bfd, template_sequence[i].data & 0xffff,
		      loc + size + 2);
          if (template_sequence[i].r_type != R_ARM_NONE)
            {
              stub_reloc_idx[nrelocs] = i;
              stub_reloc_offset[nrelocs++] = size;
            }
          size += 4;
          break;
 
	case ARM_TYPE:
	  bfd_put_32 (stub_bfd, template_sequence[i].data,
		      loc + size);
	  /* Handle cases where the target is encoded within the
	     instruction.  */
	  if (template_sequence[i].r_type == R_ARM_JUMP24)
	    {
	      stub_reloc_idx[nrelocs] = i;
	      stub_reloc_offset[nrelocs++] = size;
	    }
	  size += 4;
	  break;
 
	case DATA_TYPE:
	  bfd_put_32 (stub_bfd, template_sequence[i].data, loc + size);
	  stub_reloc_idx[nrelocs] = i;
	  stub_reloc_offset[nrelocs++] = size;
	  size += 4;
	  break;
 
	default:
	  BFD_FAIL ();
	  return FALSE;
	}
    }
 
  stub_sec->size += size;
 
  /* Stub size has already been computed in arm_size_one_stub. Check
     consistency.  */
  BFD_ASSERT (size == stub_entry->stub_size);
 
  /* Destination is Thumb. Force bit 0 to 1 to reflect this.  */
  if (stub_entry->branch_type == ST_BRANCH_TO_THUMB)
    sym_value |= 1;
 
  /* Assume there is at least one and at most MAXRELOCS entries to relocate
     in each stub.  */
  BFD_ASSERT (nrelocs != 0 && nrelocs <= MAXRELOCS);
 
  for (i = 0; i < nrelocs; i++)
    if (template_sequence[stub_reloc_idx[i]].r_type == R_ARM_THM_JUMP24
	|| template_sequence[stub_reloc_idx[i]].r_type == R_ARM_THM_JUMP19
	|| template_sequence[stub_reloc_idx[i]].r_type == R_ARM_THM_CALL
	|| template_sequence[stub_reloc_idx[i]].r_type == R_ARM_THM_XPC22)
      {
	Elf_Internal_Rela rel;
	bfd_boolean unresolved_reloc;
	char *error_message;
	enum arm_st_branch_type branch_type
	  = (template_sequence[stub_reloc_idx[i]].r_type != R_ARM_THM_XPC22
	     ? ST_BRANCH_TO_THUMB : ST_BRANCH_TO_ARM);
	bfd_vma points_to = sym_value + stub_entry->target_addend;
 
	rel.r_offset = stub_entry->stub_offset + stub_reloc_offset[i];
	rel.r_info = ELF32_R_INFO (0,
                                   template_sequence[stub_reloc_idx[i]].r_type);
	rel.r_addend = template_sequence[stub_reloc_idx[i]].reloc_addend;
 
	if (stub_entry->stub_type == arm_stub_a8_veneer_b_cond && i == 0)
	  /* The first relocation in the elf32_arm_stub_a8_veneer_b_cond[]
	     template should refer back to the instruction after the original
	     branch.  */
	  points_to = sym_value;
 
	/* There may be unintended consequences if this is not true.  */
	BFD_ASSERT (stub_entry->h == NULL);
 
	/* Note: _bfd_final_link_relocate doesn't handle these relocations
	   properly.  We should probably use this function unconditionally,
	   rather than only for certain relocations listed in the enclosing
	   conditional, for the sake of consistency.  */
	elf32_arm_final_link_relocate (elf32_arm_howto_from_type
	    (template_sequence[stub_reloc_idx[i]].r_type),
	  stub_bfd, info->output_bfd, stub_sec, stub_sec->contents, &rel,
	  points_to, info, stub_entry->target_section, "", STT_FUNC,
	  branch_type, (struct elf_link_hash_entry *) stub_entry->h,
	  &unresolved_reloc, &error_message);
      }
    else
      {
	Elf_Internal_Rela rel;
	bfd_boolean unresolved_reloc;
	char *error_message;
	bfd_vma points_to = sym_value + stub_entry->target_addend
	  + template_sequence[stub_reloc_idx[i]].reloc_addend;
 
	rel.r_offset = stub_entry->stub_offset + stub_reloc_offset[i];
	rel.r_info = ELF32_R_INFO (0,
                                   template_sequence[stub_reloc_idx[i]].r_type);
	rel.r_addend = 0;
 
	elf32_arm_final_link_relocate (elf32_arm_howto_from_type
	    (template_sequence[stub_reloc_idx[i]].r_type),
	  stub_bfd, info->output_bfd, stub_sec, stub_sec->contents, &rel,
	  points_to, info, stub_entry->target_section, "", STT_FUNC,
	  stub_entry->branch_type,
	  (struct elf_link_hash_entry *) stub_entry->h, &unresolved_reloc,
	  &error_message);
      }
 
  return TRUE;
#undef MAXRELOCS
}
 
/* Calculate the template, template size and instruction size for a stub.
   Return value is the instruction size.  */
 
static unsigned int
find_stub_size_and_template (enum elf32_arm_stub_type stub_type,
			     const insn_sequence **stub_template,
			     int *stub_template_size)
{
  const insn_sequence *template_sequence = NULL;
  int template_size = 0, i;
  unsigned int size;
 
  template_sequence = stub_definitions[stub_type].template_sequence;
  if (stub_template)
    *stub_template = template_sequence;
 
  template_size = stub_definitions[stub_type].template_size;
  if (stub_template_size)
    *stub_template_size = template_size;
 
  size = 0;
  for (i = 0; i < template_size; i++)
    {
      switch (template_sequence[i].type)
	{
	case THUMB16_TYPE:
	  size += 2;
	  break;
 
	case ARM_TYPE:
	case THUMB32_TYPE:
	case DATA_TYPE:
	  size += 4;
	  break;
 
	default:
	  BFD_FAIL ();
	  return 0;
	}
    }
 
  return size;
}
 
/* As above, but don't actually build the stub.  Just bump offset so
   we know stub section sizes.  */
 
static bfd_boolean
arm_size_one_stub (struct bfd_hash_entry *gen_entry,
		   void *in_arg ATTRIBUTE_UNUSED)
{
  struct elf32_arm_stub_hash_entry *stub_entry;
  const insn_sequence *template_sequence;
  int template_size, size;
 
  /* Massage our args to the form they really have.  */
  stub_entry = (struct elf32_arm_stub_hash_entry *) gen_entry;
 
  BFD_ASSERT((stub_entry->stub_type > arm_stub_none)
	     && stub_entry->stub_type < ARRAY_SIZE(stub_definitions));
 
  size = find_stub_size_and_template (stub_entry->stub_type, &template_sequence,
				      &template_size);
 
  stub_entry->stub_size = size;
  stub_entry->stub_template = template_sequence;
  stub_entry->stub_template_size = template_size;
 
  size = (size + 7) & ~7;
  stub_entry->stub_sec->size += size;
 
  return TRUE;
}
 
/* External entry points for sizing and building linker stubs.  */
 
/* Set up various things so that we can make a list of input sections
   for each output section included in the link.  Returns -1 on error,
   0 when no stubs will be needed, and 1 on success.  */
 
int
elf32_arm_setup_section_lists (bfd *output_bfd,
			       struct bfd_link_info *info)
{
  bfd *input_bfd;
  unsigned int bfd_count;
  int top_id, top_index;
  asection *section;
  asection **input_list, **list;
  bfd_size_type amt;
  struct elf32_arm_link_hash_table *htab = elf32_arm_hash_table (info);
 
  if (htab == NULL)
    return 0;
  if (! is_elf_hash_table (htab))
    return 0;
 
  /* Count the number of input BFDs and find the top input section id.  */
  for (input_bfd = info->input_bfds, bfd_count = 0, top_id = 0;
       input_bfd != NULL;
       input_bfd = input_bfd->link_next)
    {
      bfd_count += 1;
      for (section = input_bfd->sections;
	   section != NULL;
	   section = section->next)
	{
	  if (top_id < section->id)
	    top_id = section->id;
	}
    }
  htab->bfd_count = bfd_count;
 
  amt = sizeof (struct map_stub) * (top_id + 1);
  htab->stub_group = (struct map_stub *) bfd_zmalloc (amt);
  if (htab->stub_group == NULL)
    return -1;
  htab->top_id = top_id;
 
  /* We can't use output_bfd->section_count here to find the top output
     section index as some sections may have been removed, and
     _bfd_strip_section_from_output doesn't renumber the indices.  */
  for (section = output_bfd->sections, top_index = 0;
       section != NULL;
       section = section->next)
    {
      if (top_index < section->index)
	top_index = section->index;
    }
 
  htab->top_index = top_index;
  amt = sizeof (asection *) * (top_index + 1);
  input_list = (asection **) bfd_malloc (amt);
  htab->input_list = input_list;
  if (input_list == NULL)
    return -1;
 
  /* For sections we aren't interested in, mark their entries with a
     value we can check later.  */
  list = input_list + top_index;
  do
    *list = bfd_abs_section_ptr;
  while (list-- != input_list);
 
  for (section = output_bfd->sections;
       section != NULL;
       section = section->next)
    {
      if ((section->flags & SEC_CODE) != 0)
	input_list[section->index] = NULL;
    }
 
  return 1;
}
 
/* The linker repeatedly calls this function for each input section,
   in the order that input sections are linked into output sections.
   Build lists of input sections to determine groupings between which
   we may insert linker stubs.  */
 
void
elf32_arm_next_input_section (struct bfd_link_info *info,
			      asection *isec)
{
  struct elf32_arm_link_hash_table *htab = elf32_arm_hash_table (info);
 
  if (htab == NULL)
    return;
 
  if (isec->output_section->index <= htab->top_index)
    {
      asection **list = htab->input_list + isec->output_section->index;
 
      if (*list != bfd_abs_section_ptr && (isec->flags & SEC_CODE) != 0)
	{
	  /* Steal the link_sec pointer for our list.  */
#define PREV_SEC(sec) (htab->stub_group[(sec)->id].link_sec)
	  /* This happens to make the list in reverse order,
	     which we reverse later.  */
	  PREV_SEC (isec) = *list;
	  *list = isec;
	}
    }
}
 
/* See whether we can group stub sections together.  Grouping stub
   sections may result in fewer stubs.  More importantly, we need to
   put all .init* and .fini* stubs at the end of the .init or
   .fini output sections respectively, because glibc splits the
   _init and _fini functions into multiple parts.  Putting a stub in
   the middle of a function is not a good idea.  */
 
static void
group_sections (struct elf32_arm_link_hash_table *htab,
		bfd_size_type stub_group_size,
		bfd_boolean stubs_always_after_branch)
{
  asection **list = htab->input_list;
 
  do
    {
      asection *tail = *list;
      asection *head;
 
      if (tail == bfd_abs_section_ptr)
	continue;
 
      /* Reverse the list: we must avoid placing stubs at the
	 beginning of the section because the beginning of the text
	 section may be required for an interrupt vector in bare metal
	 code.  */
#define NEXT_SEC PREV_SEC
      head = NULL;
      while (tail != NULL)
        {
          /* Pop from tail.  */
          asection *item = tail;
          tail = PREV_SEC (item);
 
          /* Push on head.  */
          NEXT_SEC (item) = head;
          head = item;
        }
 
      while (head != NULL)
	{
	  asection *curr;
	  asection *next;
	  bfd_vma stub_group_start = head->output_offset;
	  bfd_vma end_of_next;
 
	  curr = head;
	  while (NEXT_SEC (curr) != NULL)
	    {
	      next = NEXT_SEC (curr);
	      end_of_next = next->output_offset + next->size;
	      if (end_of_next - stub_group_start >= stub_group_size)
		/* End of NEXT is too far from start, so stop.  */
		break;
	      /* Add NEXT to the group.  */
	      curr = next;
	    }
 
	  /* OK, the size from the start to the start of CURR is less
	     than stub_group_size and thus can be handled by one stub
	     section.  (Or the head section is itself larger than
	     stub_group_size, in which case we may be toast.)
	     We should really be keeping track of the total size of
	     stubs added here, as stubs contribute to the final output
	     section size.  */
	  do
	    {
	      next = NEXT_SEC (head);
	      /* Set up this stub group.  */
	      htab->stub_group[head->id].link_sec = curr;
	    }
	  while (head != curr && (head = next) != NULL);
 
	  /* But wait, there's more!  Input sections up to stub_group_size
	     bytes after the stub section can be handled by it too.  */
	  if (!stubs_always_after_branch)
	    {
	      stub_group_start = curr->output_offset + curr->size;
 
	      while (next != NULL)
		{
		  end_of_next = next->output_offset + next->size;
		  if (end_of_next - stub_group_start >= stub_group_size)
		    /* End of NEXT is too far from stubs, so stop.  */
		    break;
		  /* Add NEXT to the stub group.  */
		  head = next;
		  next = NEXT_SEC (head);
		  htab->stub_group[head->id].link_sec = curr;
		}
	    }
	  head = next;
	}
    }
  while (list++ != htab->input_list + htab->top_index);
 
  free (htab->input_list);
#undef PREV_SEC
#undef NEXT_SEC
}
 
/* Comparison function for sorting/searching relocations relating to Cortex-A8
   erratum fix.  */
 
static int
a8_reloc_compare (const void *a, const void *b)
{
  const struct a8_erratum_reloc *ra = (const struct a8_erratum_reloc *) a;
  const struct a8_erratum_reloc *rb = (const struct a8_erratum_reloc *) b;
 
  if (ra->from < rb->from)
    return -1;
  else if (ra->from > rb->from)
    return 1;
  else
    return 0;
}
 
static struct elf_link_hash_entry *find_thumb_glue (struct bfd_link_info *,
						    const char *, char **);
 
/* Helper function to scan code for sequences which might trigger the Cortex-A8
   branch/TLB erratum.  Fill in the table described by A8_FIXES_P,
   NUM_A8_FIXES_P, A8_FIX_TABLE_SIZE_P.  Returns true if an error occurs, false
   otherwise.  */
 
static bfd_boolean
cortex_a8_erratum_scan (bfd *input_bfd,
			struct bfd_link_info *info,
			struct a8_erratum_fix **a8_fixes_p,
			unsigned int *num_a8_fixes_p,
			unsigned int *a8_fix_table_size_p,
			struct a8_erratum_reloc *a8_relocs,
			unsigned int num_a8_relocs,
			unsigned prev_num_a8_fixes,
			bfd_boolean *stub_changed_p)
{
  asection *section;
  struct elf32_arm_link_hash_table *htab = elf32_arm_hash_table (info);
  struct a8_erratum_fix *a8_fixes = *a8_fixes_p;
  unsigned int num_a8_fixes = *num_a8_fixes_p;
  unsigned int a8_fix_table_size = *a8_fix_table_size_p;
 
  if (htab == NULL)
    return FALSE;
 
  for (section = input_bfd->sections;
       section != NULL;
       section = section->next)
    {
      bfd_byte *contents = NULL;
      struct _arm_elf_section_data *sec_data;
      unsigned int span;
      bfd_vma base_vma;
 
      if (elf_section_type (section) != SHT_PROGBITS
          || (elf_section_flags (section) & SHF_EXECINSTR) == 0
          || (section->flags & SEC_EXCLUDE) != 0
          || (section->sec_info_type == ELF_INFO_TYPE_JUST_SYMS)
          || (section->output_section == bfd_abs_section_ptr))
        continue;
 
      base_vma = section->output_section->vma + section->output_offset;
 
      if (elf_section_data (section)->this_hdr.contents != NULL)
        contents = elf_section_data (section)->this_hdr.contents;
      else if (! bfd_malloc_and_get_section (input_bfd, section, &contents))
        return TRUE;
 
      sec_data = elf32_arm_section_data (section);
 
      for (span = 0; span < sec_data->mapcount; span++)
        {
          unsigned int span_start = sec_data->map[span].vma;
          unsigned int span_end = (span == sec_data->mapcount - 1)
            ? section->size : sec_data->map[span + 1].vma;
          unsigned int i;
          char span_type = sec_data->map[span].type;
          bfd_boolean last_was_32bit = FALSE, last_was_branch = FALSE;
 
          if (span_type != 't')
            continue;
 
          /* Span is entirely within a single 4KB region: skip scanning.  */
          if (((base_vma + span_start) & ~0xfff)
	      == ((base_vma + span_end) & ~0xfff))
            continue;
 
          /* Scan for 32-bit Thumb-2 branches which span two 4K regions, where:
 
               * The opcode is BLX.W, BL.W, B.W, Bcc.W
               * The branch target is in the same 4KB region as the
                 first half of the branch.
               * The instruction before the branch is a 32-bit
                 length non-branch instruction.  */
          for (i = span_start; i < span_end;)
            {
              unsigned int insn = bfd_getl16 (&contents[i]);
              bfd_boolean insn_32bit = FALSE, is_blx = FALSE, is_b = FALSE;
	      bfd_boolean is_bl = FALSE, is_bcc = FALSE, is_32bit_branch;
 
              if ((insn & 0xe000) == 0xe000 && (insn & 0x1800) != 0x0000)
                insn_32bit = TRUE;
 
	      if (insn_32bit)
	        {
                  /* Load the rest of the insn (in manual-friendly order).  */
                  insn = (insn << 16) | bfd_getl16 (&contents[i + 2]);
 
        	  /* Encoding T4: B<c>.W.  */
        	  is_b = (insn & 0xf800d000) == 0xf0009000;
        	  /* Encoding T1: BL<c>.W.  */
        	  is_bl = (insn & 0xf800d000) == 0xf000d000;
        	  /* Encoding T2: BLX<c>.W.  */
        	  is_blx = (insn & 0xf800d000) == 0xf000c000;
		  /* Encoding T3: B<c>.W (not permitted in IT block).  */
		  is_bcc = (insn & 0xf800d000) == 0xf0008000
			   && (insn & 0x07f00000) != 0x03800000;
		}
 
	      is_32bit_branch = is_b || is_bl || is_blx || is_bcc;
 
              if (((base_vma + i) & 0xfff) == 0xffe
		  && insn_32bit
		  && is_32bit_branch
		  && last_was_32bit
		  && ! last_was_branch)
                {
                  bfd_signed_vma offset = 0;
                  bfd_boolean force_target_arm = FALSE;
		  bfd_boolean force_target_thumb = FALSE;
                  bfd_vma target;
                  enum elf32_arm_stub_type stub_type = arm_stub_none;
                  struct a8_erratum_reloc key, *found;
                  bfd_boolean use_plt = FALSE;
 
                  key.from = base_vma + i;
                  found = (struct a8_erratum_reloc *)
                      bsearch (&key, a8_relocs, num_a8_relocs,
                               sizeof (struct a8_erratum_reloc),
                               &a8_reloc_compare);
 
		  if (found)
		    {
		      char *error_message = NULL;
		      struct elf_link_hash_entry *entry;
 
		      /* We don't care about the error returned from this
		         function, only if there is glue or not.  */
		      entry = find_thumb_glue (info, found->sym_name,
					       &error_message);
 
		      if (entry)
			found->non_a8_stub = TRUE;
 
		      /* Keep a simpler condition, for the sake of clarity.  */
		      if (htab->root.splt != NULL && found->hash != NULL
			  && found->hash->root.plt.offset != (bfd_vma) -1)
			use_plt = TRUE;
 
		      if (found->r_type == R_ARM_THM_CALL)
			{
			  if (found->branch_type == ST_BRANCH_TO_ARM
			      || use_plt)
			    force_target_arm = TRUE;
			  else
			    force_target_thumb = TRUE;
			}
		    }
 
                  /* Check if we have an offending branch instruction.  */
 
		  if (found && found->non_a8_stub)
		    /* We've already made a stub for this instruction, e.g.
		       it's a long branch or a Thumb->ARM stub.  Assume that
		       stub will suffice to work around the A8 erratum (see
		       setting of always_after_branch above).  */
		    ;
                  else if (is_bcc)
                    {
                      offset = (insn & 0x7ff) << 1;
                      offset |= (insn & 0x3f0000) >> 4;
                      offset |= (insn & 0x2000) ? 0x40000 : 0;
                      offset |= (insn & 0x800) ? 0x80000 : 0;
                      offset |= (insn & 0x4000000) ? 0x100000 : 0;
                      if (offset & 0x100000)
                        offset |= ~ ((bfd_signed_vma) 0xfffff);
                      stub_type = arm_stub_a8_veneer_b_cond;
                    }
                  else if (is_b || is_bl || is_blx)
                    {
                      int s = (insn & 0x4000000) != 0;
                      int j1 = (insn & 0x2000) != 0;
                      int j2 = (insn & 0x800) != 0;
                      int i1 = !(j1 ^ s);
                      int i2 = !(j2 ^ s);
 
                      offset = (insn & 0x7ff) << 1;
                      offset |= (insn & 0x3ff0000) >> 4;
                      offset |= i2 << 22;
                      offset |= i1 << 23;
                      offset |= s << 24;
                      if (offset & 0x1000000)
                        offset |= ~ ((bfd_signed_vma) 0xffffff);
 
                      if (is_blx)
                        offset &= ~ ((bfd_signed_vma) 3);
 
                      stub_type = is_blx ? arm_stub_a8_veneer_blx :
                        is_bl ? arm_stub_a8_veneer_bl : arm_stub_a8_veneer_b;
                    }
 
                  if (stub_type != arm_stub_none)
                    {
                      bfd_vma pc_for_insn = base_vma + i + 4;
 
		      /* The original instruction is a BL, but the target is
		         an ARM instruction.  If we were not making a stub,
			 the BL would have been converted to a BLX.  Use the
			 BLX stub instead in that case.  */
		      if (htab->use_blx && force_target_arm
			  && stub_type == arm_stub_a8_veneer_bl)
			{
			  stub_type = arm_stub_a8_veneer_blx;
			  is_blx = TRUE;
			  is_bl = FALSE;
			}
		      /* Conversely, if the original instruction was
			 BLX but the target is Thumb mode, use the BL
			 stub.  */
		      else if (force_target_thumb
			       && stub_type == arm_stub_a8_veneer_blx)
			{
			  stub_type = arm_stub_a8_veneer_bl;
			  is_blx = FALSE;
			  is_bl = TRUE;
			}
 
                      if (is_blx)
                        pc_for_insn &= ~ ((bfd_vma) 3);
 
                      /* If we found a relocation, use the proper destination,
		         not the offset in the (unrelocated) instruction.
			 Note this is always done if we switched the stub type
			 above.  */
                      if (found)
                        offset =
			  (bfd_signed_vma) (found->destination - pc_for_insn);
 
                      /* If the stub will use a Thumb-mode branch to a
                         PLT target, redirect it to the preceding Thumb
                         entry point.  */
                      if (stub_type != arm_stub_a8_veneer_blx && use_plt)
                        offset -= PLT_THUMB_STUB_SIZE;
 
                      target = pc_for_insn + offset;
 
                      /* The BLX stub is ARM-mode code.  Adjust the offset to
		         take the different PC value (+8 instead of +4) into
			 account.  */
                      if (stub_type == arm_stub_a8_veneer_blx)
                        offset += 4;
 
                      if (((base_vma + i) & ~0xfff) == (target & ~0xfff))
                        {
                          char *stub_name = NULL;
 
                          if (num_a8_fixes == a8_fix_table_size)
                            {
                              a8_fix_table_size *= 2;
                              a8_fixes = (struct a8_erratum_fix *)
                                  bfd_realloc (a8_fixes,
                                               sizeof (struct a8_erratum_fix)
                                               * a8_fix_table_size);
                            }
 
			  if (num_a8_fixes < prev_num_a8_fixes)
			    {
			      /* If we're doing a subsequent scan,
				 check if we've found the same fix as
				 before, and try and reuse the stub
				 name.  */
			      stub_name = a8_fixes[num_a8_fixes].stub_name;
			      if ((a8_fixes[num_a8_fixes].section != section)
				  || (a8_fixes[num_a8_fixes].offset != i))
				{
				  free (stub_name);
				  stub_name = NULL;
				  *stub_changed_p = TRUE;
				}
			    }
 
			  if (!stub_name)
			    {
			      stub_name = (char *) bfd_malloc (8 + 1 + 8 + 1);
			      if (stub_name != NULL)
				sprintf (stub_name, "%x:%x", section->id, i);
			    }
 
                          a8_fixes[num_a8_fixes].input_bfd = input_bfd;
                          a8_fixes[num_a8_fixes].section = section;
                          a8_fixes[num_a8_fixes].offset = i;
                          a8_fixes[num_a8_fixes].addend = offset;
                          a8_fixes[num_a8_fixes].orig_insn = insn;
                          a8_fixes[num_a8_fixes].stub_name = stub_name;
                          a8_fixes[num_a8_fixes].stub_type = stub_type;
                          a8_fixes[num_a8_fixes].branch_type =
			    is_blx ? ST_BRANCH_TO_ARM : ST_BRANCH_TO_THUMB;
 
                          num_a8_fixes++;
                        }
                    }
                }
 
              i += insn_32bit ? 4 : 2;
              last_was_32bit = insn_32bit;
	      last_was_branch = is_32bit_branch;
            }
        }
 
      if (elf_section_data (section)->this_hdr.contents == NULL)
        free (contents);
    }
 
  *a8_fixes_p = a8_fixes;
  *num_a8_fixes_p = num_a8_fixes;
  *a8_fix_table_size_p = a8_fix_table_size;
 
  return FALSE;
}
 
/* Determine and set the size of the stub section for a final link.
 
   The basic idea here is to examine all the relocations looking for
   PC-relative calls to a target that is unreachable with a "bl"
   instruction.  */
 
bfd_boolean
elf32_arm_size_stubs (bfd *output_bfd,
		      bfd *stub_bfd,
		      struct bfd_link_info *info,
		      bfd_signed_vma group_size,
		      asection * (*add_stub_section) (const char *, asection *),
		      void (*layout_sections_again) (void))
{
  bfd_size_type stub_group_size;
  bfd_boolean stubs_always_after_branch;
  struct elf32_arm_link_hash_table *htab = elf32_arm_hash_table (info);
  struct a8_erratum_fix *a8_fixes = NULL;
  unsigned int num_a8_fixes = 0, a8_fix_table_size = 10;
  struct a8_erratum_reloc *a8_relocs = NULL;
  unsigned int num_a8_relocs = 0, a8_reloc_table_size = 10, i;
 
  if (htab == NULL)
    return FALSE;
 
  if (htab->fix_cortex_a8)
    {
      a8_fixes = (struct a8_erratum_fix *)
          bfd_zmalloc (sizeof (struct a8_erratum_fix) * a8_fix_table_size);
      a8_relocs = (struct a8_erratum_reloc *)
          bfd_zmalloc (sizeof (struct a8_erratum_reloc) * a8_reloc_table_size);
    }
 
  /* Propagate mach to stub bfd, because it may not have been
     finalized when we created stub_bfd.  */
  bfd_set_arch_mach (stub_bfd, bfd_get_arch (output_bfd),
		     bfd_get_mach (output_bfd));
 
  /* Stash our params away.  */
  htab->stub_bfd = stub_bfd;
  htab->add_stub_section = add_stub_section;
  htab->layout_sections_again = layout_sections_again;
  stubs_always_after_branch = group_size < 0;
 
  /* The Cortex-A8 erratum fix depends on stubs not being in the same 4K page
     as the first half of a 32-bit branch straddling two 4K pages.  This is a
     crude way of enforcing that.  */
  if (htab->fix_cortex_a8)
    stubs_always_after_branch = 1;
 
  if (group_size < 0)
    stub_group_size = -group_size;
  else
    stub_group_size = group_size;
 
  if (stub_group_size == 1)
    {
      /* Default values.  */
      /* Thumb branch range is +-4MB has to be used as the default
	 maximum size (a given section can contain both ARM and Thumb
	 code, so the worst case has to be taken into account).
 
	 This value is 24K less than that, which allows for 2025
	 12-byte stubs.  If we exceed that, then we will fail to link.
	 The user will have to relink with an explicit group size
	 option.  */
      stub_group_size = 4170000;
    }
 
  group_sections (htab, stub_group_size, stubs_always_after_branch);
 
  /* If we're applying the cortex A8 fix, we need to determine the
     program header size now, because we cannot change it later --
     that could alter section placements.  Notice the A8 erratum fix
     ends up requiring the section addresses to remain unchanged
     modulo the page size.  That's something we cannot represent
     inside BFD, and we don't want to force the section alignment to
     be the page size.  */
  if (htab->fix_cortex_a8)
    (*htab->layout_sections_again) ();
 
  while (1)
    {
      bfd *input_bfd;
      unsigned int bfd_indx;
      asection *stub_sec;
      bfd_boolean stub_changed = FALSE;
      unsigned prev_num_a8_fixes = num_a8_fixes;
 
      num_a8_fixes = 0;
      for (input_bfd = info->input_bfds, bfd_indx = 0;
	   input_bfd != NULL;
	   input_bfd = input_bfd->link_next, bfd_indx++)
	{
	  Elf_Internal_Shdr *symtab_hdr;
	  asection *section;
	  Elf_Internal_Sym *local_syms = NULL;
 
	  num_a8_relocs = 0;
 
	  /* We'll need the symbol table in a second.  */
	  symtab_hdr = &elf_tdata (input_bfd)->symtab_hdr;
	  if (symtab_hdr->sh_info == 0)
	    continue;
 
	  /* Walk over each section attached to the input bfd.  */
	  for (section = input_bfd->sections;
	       section != NULL;
	       section = section->next)
	    {
	      Elf_Internal_Rela *internal_relocs, *irelaend, *irela;
 
	      /* If there aren't any relocs, then there's nothing more
		 to do.  */
	      if ((section->flags & SEC_RELOC) == 0
		  || section->reloc_count == 0
		  || (section->flags & SEC_CODE) == 0)
		continue;
 
	      /* If this section is a link-once section that will be
		 discarded, then don't create any stubs.  */
	      if (section->output_section == NULL
		  || section->output_section->owner != output_bfd)
		continue;
 
	      /* Get the relocs.  */
	      internal_relocs
		= _bfd_elf_link_read_relocs (input_bfd, section, NULL,
					     NULL, info->keep_memory);
	      if (internal_relocs == NULL)
		goto error_ret_free_local;
 
	      /* Now examine each relocation.  */
	      irela = internal_relocs;
	      irelaend = irela + section->reloc_count;
	      for (; irela < irelaend; irela++)
		{
		  unsigned int r_type, r_indx;
		  enum elf32_arm_stub_type stub_type;
		  struct elf32_arm_stub_hash_entry *stub_entry;
		  asection *sym_sec;
		  bfd_vma sym_value;
		  bfd_vma destination;
		  struct elf32_arm_link_hash_entry *hash;
		  const char *sym_name;
		  char *stub_name;
		  const asection *id_sec;
		  unsigned char st_type;
		  enum arm_st_branch_type branch_type;
		  bfd_boolean created_stub = FALSE;
 
		  r_type = ELF32_R_TYPE (irela->r_info);
		  r_indx = ELF32_R_SYM (irela->r_info);
 
		  if (r_type >= (unsigned int) R_ARM_max)
		    {
		      bfd_set_error (bfd_error_bad_value);
		    error_ret_free_internal:
		      if (elf_section_data (section)->relocs == NULL)
			free (internal_relocs);
		      goto error_ret_free_local;
		    }
 
		  hash = NULL;
		  if (r_indx >= symtab_hdr->sh_info)
		    hash = elf32_arm_hash_entry
		      (elf_sym_hashes (input_bfd)
		       [r_indx - symtab_hdr->sh_info]);
 
		  /* Only look for stubs on branch instructions, or
		     non-relaxed TLSCALL  */
		  if ((r_type != (unsigned int) R_ARM_CALL)
		      && (r_type != (unsigned int) R_ARM_THM_CALL)
		      && (r_type != (unsigned int) R_ARM_JUMP24)
		      && (r_type != (unsigned int) R_ARM_THM_JUMP19)
		      && (r_type != (unsigned int) R_ARM_THM_XPC22)
		      && (r_type != (unsigned int) R_ARM_THM_JUMP24)
		      && (r_type != (unsigned int) R_ARM_PLT32)
		      && !((r_type == (unsigned int) R_ARM_TLS_CALL
			    || r_type == (unsigned int) R_ARM_THM_TLS_CALL)
			   && r_type == elf32_arm_tls_transition
			       (info, r_type, &hash->root)
			   && ((hash ? hash->tls_type
				: (elf32_arm_local_got_tls_type
				   (input_bfd)[r_indx]))
			       & GOT_TLS_GDESC) != 0))
		    continue;
 
		  /* Now determine the call target, its name, value,
		     section.  */
		  sym_sec = NULL;
		  sym_value = 0;
		  destination = 0;
		  sym_name = NULL;
 
		  if (r_type == (unsigned int) R_ARM_TLS_CALL
		      || r_type == (unsigned int) R_ARM_THM_TLS_CALL)
		    {
		      /* A non-relaxed TLS call.  The target is the
			 plt-resident trampoline and nothing to do
			 with the symbol.  */
		      BFD_ASSERT (htab->tls_trampoline > 0);
		      sym_sec = htab->root.splt;
		      sym_value = htab->tls_trampoline;
		      hash = 0;
		      st_type = STT_FUNC;
		      branch_type = ST_BRANCH_TO_ARM;
		    }
		  else if (!hash)
		    {
		      /* It's a local symbol.  */
		      Elf_Internal_Sym *sym;
 
		      if (local_syms == NULL)
			{
			  local_syms
			    = (Elf_Internal_Sym *) symtab_hdr->contents;
			  if (local_syms == NULL)
			    local_syms
			      = bfd_elf_get_elf_syms (input_bfd, symtab_hdr,
						      symtab_hdr->sh_info, 0,
						      NULL, NULL, NULL);
			  if (local_syms == NULL)
			    goto error_ret_free_internal;
			}
 
		      sym = local_syms + r_indx;
		      if (sym->st_shndx == SHN_UNDEF)
			sym_sec = bfd_und_section_ptr;
		      else if (sym->st_shndx == SHN_ABS)
			sym_sec = bfd_abs_section_ptr;
		      else if (sym->st_shndx == SHN_COMMON)
			sym_sec = bfd_com_section_ptr;
		      else
			sym_sec =
			  bfd_section_from_elf_index (input_bfd, sym->st_shndx);
 
		      if (!sym_sec)
			/* This is an undefined symbol.  It can never
			   be resolved. */
			continue;
 
		      if (ELF_ST_TYPE (sym->st_info) != STT_SECTION)
			sym_value = sym->st_value;
		      destination = (sym_value + irela->r_addend
				     + sym_sec->output_offset
				     + sym_sec->output_section->vma);
		      st_type = ELF_ST_TYPE (sym->st_info);
		      branch_type = ARM_SYM_BRANCH_TYPE (sym);
		      sym_name
			= bfd_elf_string_from_elf_section (input_bfd,
							   symtab_hdr->sh_link,
							   sym->st_name);
		    }
		  else
		    {
		      /* It's an external symbol.  */
		      while (hash->root.root.type == bfd_link_hash_indirect
			     || hash->root.root.type == bfd_link_hash_warning)
			hash = ((struct elf32_arm_link_hash_entry *)
				hash->root.root.u.i.link);
 
		      if (hash->root.root.type == bfd_link_hash_defined
			  || hash->root.root.type == bfd_link_hash_defweak)
			{
			  sym_sec = hash->root.root.u.def.section;
			  sym_value = hash->root.root.u.def.value;
 
			  struct elf32_arm_link_hash_table *globals =
						  elf32_arm_hash_table (info);
 
			  /* For a destination in a shared library,
			     use the PLT stub as target address to
			     decide whether a branch stub is
			     needed.  */
			  if (globals != NULL
			      && globals->root.splt != NULL
			      && hash != NULL
			      && hash->root.plt.offset != (bfd_vma) -1)
			    {
			      sym_sec = globals->root.splt;
			      sym_value = hash->root.plt.offset;
			      if (sym_sec->output_section != NULL)
				destination = (sym_value
					       + sym_sec->output_offset
					       + sym_sec->output_section->vma);
			    }
			  else if (sym_sec->output_section != NULL)
			    destination = (sym_value + irela->r_addend
					   + sym_sec->output_offset
					   + sym_sec->output_section->vma);
			}
		      else if ((hash->root.root.type == bfd_link_hash_undefined)
			       || (hash->root.root.type == bfd_link_hash_undefweak))
			{
			  /* For a shared library, use the PLT stub as
			     target address to decide whether a long
			     branch stub is needed.
			     For absolute code, they cannot be handled.  */
			  struct elf32_arm_link_hash_table *globals =
			    elf32_arm_hash_table (info);
 
			  if (globals != NULL
			      && globals->root.splt != NULL
			      && hash != NULL
			      && hash->root.plt.offset != (bfd_vma) -1)
			    {
			      sym_sec = globals->root.splt;
			      sym_value = hash->root.plt.offset;
			      if (sym_sec->output_section != NULL)
				destination = (sym_value
					       + sym_sec->output_offset
					       + sym_sec->output_section->vma);
			    }
			  else
			    continue;
			}
		      else
			{
			  bfd_set_error (bfd_error_bad_value);
			  goto error_ret_free_internal;
			}
		      st_type = hash->root.type;
		      branch_type = hash->root.target_internal;
		      sym_name = hash->root.root.root.string;
		    }
 
		  do
		    {
		      /* Determine what (if any) linker stub is needed.  */
		      stub_type = arm_type_of_stub (info, section, irela,
						    st_type, &branch_type,
						    hash, destination, sym_sec,
						    input_bfd, sym_name);
		      if (stub_type == arm_stub_none)
			break;
 
		      /* Support for grouping stub sections.  */
		      id_sec = htab->stub_group[section->id].link_sec;
 
		      /* Get the name of this stub.  */
		      stub_name = elf32_arm_stub_name (id_sec, sym_sec, hash,
						       irela, stub_type);
		      if (!stub_name)
			goto error_ret_free_internal;
 
		      /* We've either created a stub for this reloc already,
			 or we are about to.  */
		      created_stub = TRUE;
 
		      stub_entry = arm_stub_hash_lookup
				     (&htab->stub_hash_table, stub_name,
				      FALSE, FALSE);
		      if (stub_entry != NULL)
			{
			  /* The proper stub has already been created.  */
			  free (stub_name);
			  stub_entry->target_value = sym_value;
			  break;
			}
 
		      stub_entry = elf32_arm_add_stub (stub_name, section,
						       htab);
		      if (stub_entry == NULL)
			{
			  free (stub_name);
			  goto error_ret_free_internal;
			}
 
                      stub_entry->target_value = sym_value;
                      stub_entry->target_section = sym_sec;
                      stub_entry->stub_type = stub_type;
                      stub_entry->h = hash;
                      stub_entry->branch_type = branch_type;
 
                      if (sym_name == NULL)
                	sym_name = "unnamed";
                      stub_entry->output_name = (char *)
                          bfd_alloc (htab->stub_bfd,
                                     sizeof (THUMB2ARM_GLUE_ENTRY_NAME)
                                     + strlen (sym_name));
                      if (stub_entry->output_name == NULL)
                	{
                          free (stub_name);
                          goto error_ret_free_internal;
                	}
 
                      /* For historical reasons, use the existing names for
                	 ARM-to-Thumb and Thumb-to-ARM stubs.  */
                      if ((r_type == (unsigned int) R_ARM_THM_CALL
			   || r_type == (unsigned int) R_ARM_THM_JUMP24)
			  && branch_type == ST_BRANCH_TO_ARM)
                	sprintf (stub_entry->output_name,
                        	 THUMB2ARM_GLUE_ENTRY_NAME, sym_name);
                      else if ((r_type == (unsigned int) R_ARM_CALL
			       || r_type == (unsigned int) R_ARM_JUMP24)
			       && branch_type == ST_BRANCH_TO_THUMB)
                	sprintf (stub_entry->output_name,
                        	 ARM2THUMB_GLUE_ENTRY_NAME, sym_name);
                      else
                	sprintf (stub_entry->output_name, STUB_ENTRY_NAME,
                        	 sym_name);
 
                      stub_changed = TRUE;
                    }
                  while (0);
 
                  /* Look for relocations which might trigger Cortex-A8
                     erratum.  */
                  if (htab->fix_cortex_a8
                      && (r_type == (unsigned int) R_ARM_THM_JUMP24
                          || r_type == (unsigned int) R_ARM_THM_JUMP19
                          || r_type == (unsigned int) R_ARM_THM_CALL
                          || r_type == (unsigned int) R_ARM_THM_XPC22))
                    {
                      bfd_vma from = section->output_section->vma
                                     + section->output_offset
                                     + irela->r_offset;
 
                      if ((from & 0xfff) == 0xffe)
                        {
                          /* Found a candidate.  Note we haven't checked the
                             destination is within 4K here: if we do so (and
                             don't create an entry in a8_relocs) we can't tell
                             that a branch should have been relocated when
                             scanning later.  */
                          if (num_a8_relocs == a8_reloc_table_size)
                            {
                              a8_reloc_table_size *= 2;
                              a8_relocs = (struct a8_erratum_reloc *)
                                  bfd_realloc (a8_relocs,
                                               sizeof (struct a8_erratum_reloc)
                                               * a8_reloc_table_size);
                            }
 
                          a8_relocs[num_a8_relocs].from = from;
                          a8_relocs[num_a8_relocs].destination = destination;
                          a8_relocs[num_a8_relocs].r_type = r_type;
                          a8_relocs[num_a8_relocs].branch_type = branch_type;
                          a8_relocs[num_a8_relocs].sym_name = sym_name;
                          a8_relocs[num_a8_relocs].non_a8_stub = created_stub;
                          a8_relocs[num_a8_relocs].hash = hash;
 
                          num_a8_relocs++;
                        }
                    }
		}
 
              /* We're done with the internal relocs, free them.  */
              if (elf_section_data (section)->relocs == NULL)
                free (internal_relocs);
            }
 
          if (htab->fix_cortex_a8)
	    {
              /* Sort relocs which might apply to Cortex-A8 erratum.  */
              qsort (a8_relocs, num_a8_relocs,
		     sizeof (struct a8_erratum_reloc),
                     &a8_reloc_compare);
 
              /* Scan for branches which might trigger Cortex-A8 erratum.  */
              if (cortex_a8_erratum_scan (input_bfd, info, &a8_fixes,
					  &num_a8_fixes, &a8_fix_table_size,
					  a8_relocs, num_a8_relocs,
					  prev_num_a8_fixes, &stub_changed)
		  != 0)
		goto error_ret_free_local;
	    }
	}
 
      if (prev_num_a8_fixes != num_a8_fixes)
        stub_changed = TRUE;
 
      if (!stub_changed)
	break;
 
      /* OK, we've added some stubs.  Find out the new size of the
	 stub sections.  */
      for (stub_sec = htab->stub_bfd->sections;
	   stub_sec != NULL;
	   stub_sec = stub_sec->next)
	{
	  /* Ignore non-stub sections.  */
	  if (!strstr (stub_sec->name, STUB_SUFFIX))
	    continue;
 
	  stub_sec->size = 0;
	}
 
      bfd_hash_traverse (&htab->stub_hash_table, arm_size_one_stub, htab);
 
      /* Add Cortex-A8 erratum veneers to stub section sizes too.  */
      if (htab->fix_cortex_a8)
        for (i = 0; i < num_a8_fixes; i++)
          {
	    stub_sec = elf32_arm_create_or_find_stub_sec (NULL,
			 a8_fixes[i].section, htab);
 
	    if (stub_sec == NULL)
	      goto error_ret_free_local;
 
            stub_sec->size
              += find_stub_size_and_template (a8_fixes[i].stub_type, NULL,
                                              NULL);
          }
 
 
      /* Ask the linker to do its stuff.  */
      (*htab->layout_sections_again) ();
    }
 
  /* Add stubs for Cortex-A8 erratum fixes now.  */
  if (htab->fix_cortex_a8)
    {
      for (i = 0; i < num_a8_fixes; i++)
        {
          struct elf32_arm_stub_hash_entry *stub_entry;
          char *stub_name = a8_fixes[i].stub_name;
          asection *section = a8_fixes[i].section;
          unsigned int section_id = a8_fixes[i].section->id;
          asection *link_sec = htab->stub_group[section_id].link_sec;
          asection *stub_sec = htab->stub_group[section_id].stub_sec;
          const insn_sequence *template_sequence;
          int template_size, size = 0;
 
          stub_entry = arm_stub_hash_lookup (&htab->stub_hash_table, stub_name,
                                             TRUE, FALSE);
          if (stub_entry == NULL)
            {
              (*_bfd_error_handler) (_("%s: cannot create stub entry %s"),
                                     section->owner,
                                     stub_name);
              return FALSE;
            }
 
          stub_entry->stub_sec = stub_sec;
          stub_entry->stub_offset = 0;
          stub_entry->id_sec = link_sec;
          stub_entry->stub_type = a8_fixes[i].stub_type;
          stub_entry->target_section = a8_fixes[i].section;
          stub_entry->target_value = a8_fixes[i].offset;
          stub_entry->target_addend = a8_fixes[i].addend;
          stub_entry->orig_insn = a8_fixes[i].orig_insn;
	  stub_entry->branch_type = a8_fixes[i].branch_type;
 
          size = find_stub_size_and_template (a8_fixes[i].stub_type,
                                              &template_sequence,
                                              &template_size);
 
          stub_entry->stub_size = size;
          stub_entry->stub_template = template_sequence;
          stub_entry->stub_template_size = template_size;
        }
 
      /* Stash the Cortex-A8 erratum fix array for use later in
         elf32_arm_write_section().  */
      htab->a8_erratum_fixes = a8_fixes;
      htab->num_a8_erratum_fixes = num_a8_fixes;
    }
  else
    {
      htab->a8_erratum_fixes = NULL;
      htab->num_a8_erratum_fixes = 0;
    }
  return TRUE;
 
 error_ret_free_local:
  return FALSE;
}
 
/* Build all the stubs associated with the current output file.  The
   stubs are kept in a hash table attached to the main linker hash
   table.  We also set up the .plt entries for statically linked PIC
   functions here.  This function is called via arm_elf_finish in the
   linker.  */
 
bfd_boolean
elf32_arm_build_stubs (struct bfd_link_info *info)
{
  asection *stub_sec;
  struct bfd_hash_table *table;
  struct elf32_arm_link_hash_table *htab;
 
  htab = elf32_arm_hash_table (info);
  if (htab == NULL)
    return FALSE;
 
  for (stub_sec = htab->stub_bfd->sections;
       stub_sec != NULL;
       stub_sec = stub_sec->next)
    {
      bfd_size_type size;
 
      /* Ignore non-stub sections.  */
      if (!strstr (stub_sec->name, STUB_SUFFIX))
	continue;
 
      /* Allocate memory to hold the linker stubs.  */
      size = stub_sec->size;
      stub_sec->contents = (unsigned char *) bfd_zalloc (htab->stub_bfd, size);
      if (stub_sec->contents == NULL && size != 0)
	return FALSE;
      stub_sec->size = 0;
    }
 
  /* Build the stubs as directed by the stub hash table.  */
  table = &htab->stub_hash_table;
  bfd_hash_traverse (table, arm_build_one_stub, info);
  if (htab->fix_cortex_a8)
    {
      /* Place the cortex a8 stubs last.  */
      htab->fix_cortex_a8 = -1;
      bfd_hash_traverse (table, arm_build_one_stub, info);
    }
 
  return TRUE;
}
 
/* Locate the Thumb encoded calling stub for NAME.  */
 
static struct elf_link_hash_entry *
find_thumb_glue (struct bfd_link_info *link_info,
		 const char *name,
		 char **error_message)
{
  char *tmp_name;
  struct elf_link_hash_entry *hash;
  struct elf32_arm_link_hash_table *hash_table;
 
  /* We need a pointer to the armelf specific hash table.  */
  hash_table = elf32_arm_hash_table (link_info);
  if (hash_table == NULL)
    return NULL;
 
  tmp_name = (char *) bfd_malloc ((bfd_size_type) strlen (name)
                                  + strlen (THUMB2ARM_GLUE_ENTRY_NAME) + 1);
 
  BFD_ASSERT (tmp_name);
 
  sprintf (tmp_name, THUMB2ARM_GLUE_ENTRY_NAME, name);
 
  hash = elf_link_hash_lookup
    (&(hash_table)->root, tmp_name, FALSE, FALSE, TRUE);
 
  if (hash == NULL
      && asprintf (error_message, _("unable to find THUMB glue '%s' for '%s'"),
		   tmp_name, name) == -1)
    *error_message = (char *) bfd_errmsg (bfd_error_system_call);
 
  free (tmp_name);
 
  return hash;
}
 
/* Locate the ARM encoded calling stub for NAME.  */
 
static struct elf_link_hash_entry *
find_arm_glue (struct bfd_link_info *link_info,
	       const char *name,
	       char **error_message)
{
  char *tmp_name;
  struct elf_link_hash_entry *myh;
  struct elf32_arm_link_hash_table *hash_table;
 
  /* We need a pointer to the elfarm specific hash table.  */
  hash_table = elf32_arm_hash_table (link_info);
  if (hash_table == NULL)
    return NULL;
 
  tmp_name = (char *) bfd_malloc ((bfd_size_type) strlen (name)
                                  + strlen (ARM2THUMB_GLUE_ENTRY_NAME) + 1);
 
  BFD_ASSERT (tmp_name);
 
  sprintf (tmp_name, ARM2THUMB_GLUE_ENTRY_NAME, name);
 
  myh = elf_link_hash_lookup
    (&(hash_table)->root, tmp_name, FALSE, FALSE, TRUE);
 
  if (myh == NULL
      && asprintf (error_message, _("unable to find ARM glue '%s' for '%s'"),
		   tmp_name, name) == -1)
    *error_message = (char *) bfd_errmsg (bfd_error_system_call);
 
  free (tmp_name);
 
  return myh;
}
 
/* ARM->Thumb glue (static images):
 
   .arm
   __func_from_arm:
   ldr r12, __func_addr
   bx  r12
   __func_addr:
   .word func    @ behave as if you saw a ARM_32 reloc.
 
   (v5t static images)
   .arm
   __func_from_arm:
   ldr pc, __func_addr
   __func_addr:
   .word func    @ behave as if you saw a ARM_32 reloc.
 
   (relocatable images)
   .arm
   __func_from_arm:
   ldr r12, __func_offset
   add r12, r12, pc
   bx  r12
   __func_offset:
   .word func - .   */
 
#define ARM2THUMB_STATIC_GLUE_SIZE 12
static const insn32 a2t1_ldr_insn = 0xe59fc000;
static const insn32 a2t2_bx_r12_insn = 0xe12fff1c;
static const insn32 a2t3_func_addr_insn = 0x00000001;
 
#define ARM2THUMB_V5_STATIC_GLUE_SIZE 8
static const insn32 a2t1v5_ldr_insn = 0xe51ff004;
static const insn32 a2t2v5_func_addr_insn = 0x00000001;
 
#define ARM2THUMB_PIC_GLUE_SIZE 16
static const insn32 a2t1p_ldr_insn = 0xe59fc004;
static const insn32 a2t2p_add_pc_insn = 0xe08cc00f;
static const insn32 a2t3p_bx_r12_insn = 0xe12fff1c;
 
/* Thumb->ARM:                          Thumb->(non-interworking aware) ARM
 
     .thumb                             .thumb
     .align 2                           .align 2
 __func_from_thumb:                 __func_from_thumb:
     bx pc                              push {r6, lr}
     nop                                ldr  r6, __func_addr
     .arm                               mov  lr, pc
     b func                             bx   r6
                                        .arm
		 		    ;; back_to_thumb       
                 		        ldmia r13! {r6, lr}
 				        bx    lr           
                                    __func_addr:
                                        .word        func  */
 
#define THUMB2ARM_GLUE_SIZE 8
static const insn16 t2a1_bx_pc_insn = 0x4778;
static const insn16 t2a2_noop_insn = 0x46c0;
static const insn32 t2a3_b_insn = 0xea000000;
 
#define VFP11_ERRATUM_VENEER_SIZE 8
 
#define ARM_BX_VENEER_SIZE 12
static const insn32 armbx1_tst_insn = 0xe3100001;
static const insn32 armbx2_moveq_insn = 0x01a0f000;
static const insn32 armbx3_bx_insn = 0xe12fff10;
 
#ifndef ELFARM_NABI_C_INCLUDED
static void
arm_allocate_glue_section_space (bfd * abfd, bfd_size_type size, const char * name)
{
  asection * s;
  bfd_byte * contents;
 
  if (size == 0)
    {
      /* Do not include empty glue sections in the output.  */
      if (abfd != NULL)
	{
	  s = bfd_get_section_by_name (abfd, name);
	  if (s != NULL)
	    s->flags |= SEC_EXCLUDE;
	}
      return;
    }
 
  BFD_ASSERT (abfd != NULL);
 
  s = bfd_get_section_by_name (abfd, name);
  BFD_ASSERT (s != NULL);
 
  contents = (bfd_byte *) bfd_alloc (abfd, size);
 
  BFD_ASSERT (s->size == size);
  s->contents = contents;
}
 
bfd_boolean
bfd_elf32_arm_allocate_interworking_sections (struct bfd_link_info * info)
{
  struct elf32_arm_link_hash_table * globals;
 
  globals = elf32_arm_hash_table (info);
  BFD_ASSERT (globals != NULL);
 
  arm_allocate_glue_section_space (globals->bfd_of_glue_owner,
				   globals->arm_glue_size,
				   ARM2THUMB_GLUE_SECTION_NAME);
 
  arm_allocate_glue_section_space (globals->bfd_of_glue_owner,
				   globals->thumb_glue_size,
				   THUMB2ARM_GLUE_SECTION_NAME);
 
  arm_allocate_glue_section_space (globals->bfd_of_glue_owner,
				   globals->vfp11_erratum_glue_size,
				   VFP11_ERRATUM_VENEER_SECTION_NAME);
 
  arm_allocate_glue_section_space (globals->bfd_of_glue_owner,
				   globals->bx_glue_size,
				   ARM_BX_GLUE_SECTION_NAME);
 
  return TRUE;
}
 
/* Allocate space and symbols for calling a Thumb function from Arm mode.
   returns the symbol identifying the stub.  */
 
static struct elf_link_hash_entry *
record_arm_to_thumb_glue (struct bfd_link_info * link_info,
			  struct elf_link_hash_entry * h)
{
  const char * name = h->root.root.string;
  asection * s;
  char * tmp_name;
  struct elf_link_hash_entry * myh;
  struct bfd_link_hash_entry * bh;
  struct elf32_arm_link_hash_table * globals;
  bfd_vma val;
  bfd_size_type size;
 
  globals = elf32_arm_hash_table (link_info);
  BFD_ASSERT (globals != NULL);
  BFD_ASSERT (globals->bfd_of_glue_owner != NULL);
 
  s = bfd_get_section_by_name
    (globals->bfd_of_glue_owner, ARM2THUMB_GLUE_SECTION_NAME);
 
  BFD_ASSERT (s != NULL);
 
  tmp_name = (char *) bfd_malloc ((bfd_size_type) strlen (name)
                                  + strlen (ARM2THUMB_GLUE_ENTRY_NAME) + 1);
 
  BFD_ASSERT (tmp_name);
 
  sprintf (tmp_name, ARM2THUMB_GLUE_ENTRY_NAME, name);
 
  myh = elf_link_hash_lookup
    (&(globals)->root, tmp_name, FALSE, FALSE, TRUE);
 
  if (myh != NULL)
    {
      /* We've already seen this guy.  */
      free (tmp_name);
      return myh;
    }
 
  /* The only trick here is using hash_table->arm_glue_size as the value.
     Even though the section isn't allocated yet, this is where we will be
     putting it.  The +1 on the value marks that the stub has not been
     output yet - not that it is a Thumb function.  */
  bh = NULL;
  val = globals->arm_glue_size + 1;
  _bfd_generic_link_add_one_symbol (link_info, globals->bfd_of_glue_owner,
				    tmp_name, BSF_GLOBAL, s, val,
				    NULL, TRUE, FALSE, &bh);
 
  myh = (struct elf_link_hash_entry *) bh;
  myh->type = ELF_ST_INFO (STB_LOCAL, STT_FUNC);
  myh->forced_local = 1;
 
  free (tmp_name);
 
  if (link_info->shared || globals->root.is_relocatable_executable
      || globals->pic_veneer)
    size = ARM2THUMB_PIC_GLUE_SIZE;
  else if (globals->use_blx)
    size = ARM2THUMB_V5_STATIC_GLUE_SIZE;
  else
    size = ARM2THUMB_STATIC_GLUE_SIZE;
 
  s->size += size;
  globals->arm_glue_size += size;
 
  return myh;
}
 
/* Allocate space for ARMv4 BX veneers.  */
 
static void
record_arm_bx_glue (struct bfd_link_info * link_info, int reg)
{
  asection * s;
  struct elf32_arm_link_hash_table *globals;
  char *tmp_name;
  struct elf_link_hash_entry *myh;
  struct bfd_link_hash_entry *bh;
  bfd_vma val;
 
  /* BX PC does not need a veneer.  */
  if (reg == 15)
    return;
 
  globals = elf32_arm_hash_table (link_info);
  BFD_ASSERT (globals != NULL);
  BFD_ASSERT (globals->bfd_of_glue_owner != NULL);
 
  /* Check if this veneer has already been allocated.  */
  if (globals->bx_glue_offset[reg])
    return;
 
  s = bfd_get_section_by_name
    (globals->bfd_of_glue_owner, ARM_BX_GLUE_SECTION_NAME);
 
  BFD_ASSERT (s != NULL);
 
  /* Add symbol for veneer.  */
  tmp_name = (char *)
      bfd_malloc ((bfd_size_type) strlen (ARM_BX_GLUE_ENTRY_NAME) + 1);
 
  BFD_ASSERT (tmp_name);
 
  sprintf (tmp_name, ARM_BX_GLUE_ENTRY_NAME, reg);
 
  myh = elf_link_hash_lookup
    (&(globals)->root, tmp_name, FALSE, FALSE, FALSE);
 
  BFD_ASSERT (myh == NULL);
 
  bh = NULL;
  val = globals->bx_glue_size;
  _bfd_generic_link_add_one_symbol (link_info, globals->bfd_of_glue_owner,
                                    tmp_name, BSF_FUNCTION | BSF_LOCAL, s, val,
                                    NULL, TRUE, FALSE, &bh);
 
  myh = (struct elf_link_hash_entry *) bh;
  myh->type = ELF_ST_INFO (STB_LOCAL, STT_FUNC);
  myh->forced_local = 1;
 
  s->size += ARM_BX_VENEER_SIZE;
  globals->bx_glue_offset[reg] = globals->bx_glue_size | 2;
  globals->bx_glue_size += ARM_BX_VENEER_SIZE;
}
 
 
/* Add an entry to the code/data map for section SEC.  */
 
static void
elf32_arm_section_map_add (asection *sec, char type, bfd_vma vma)
{
  struct _arm_elf_section_data *sec_data = elf32_arm_section_data (sec);
  unsigned int newidx;
 
  if (sec_data->map == NULL)
    {
      sec_data->map = (elf32_arm_section_map *)
          bfd_malloc (sizeof (elf32_arm_section_map));
      sec_data->mapcount = 0;
      sec_data->mapsize = 1;
    }
 
  newidx = sec_data->mapcount++;
 
  if (sec_data->mapcount > sec_data->mapsize)
    {
      sec_data->mapsize *= 2;
      sec_data->map = (elf32_arm_section_map *)
          bfd_realloc_or_free (sec_data->map, sec_data->mapsize
                               * sizeof (elf32_arm_section_map));
    }
 
  if (sec_data->map)
    {
      sec_data->map[newidx].vma = vma;
      sec_data->map[newidx].type = type;
    }
}
 
 
/* Record information about a VFP11 denorm-erratum veneer.  Only ARM-mode
   veneers are handled for now.  */
 
static bfd_vma
record_vfp11_erratum_veneer (struct bfd_link_info *link_info,
                             elf32_vfp11_erratum_list *branch,
                             bfd *branch_bfd,
                             asection *branch_sec,
                             unsigned int offset)
{
  asection *s;
  struct elf32_arm_link_hash_table *hash_table;
  char *tmp_name;
  struct elf_link_hash_entry *myh;
  struct bfd_link_hash_entry *bh;
  bfd_vma val;
  struct _arm_elf_section_data *sec_data;
  elf32_vfp11_erratum_list *newerr;
 
  hash_table = elf32_arm_hash_table (link_info);
  BFD_ASSERT (hash_table != NULL);
  BFD_ASSERT (hash_table->bfd_of_glue_owner != NULL);
 
  s = bfd_get_section_by_name
    (hash_table->bfd_of_glue_owner, VFP11_ERRATUM_VENEER_SECTION_NAME);
 
  sec_data = elf32_arm_section_data (s);
 
  BFD_ASSERT (s != NULL);
 
  tmp_name = (char *) bfd_malloc ((bfd_size_type) strlen
                                  (VFP11_ERRATUM_VENEER_ENTRY_NAME) + 10);
 
  BFD_ASSERT (tmp_name);
 
  sprintf (tmp_name, VFP11_ERRATUM_VENEER_ENTRY_NAME,
	   hash_table->num_vfp11_fixes);
 
  myh = elf_link_hash_lookup
    (&(hash_table)->root, tmp_name, FALSE, FALSE, FALSE);
 
  BFD_ASSERT (myh == NULL);
 
  bh = NULL;
  val = hash_table->vfp11_erratum_glue_size;
  _bfd_generic_link_add_one_symbol (link_info, hash_table->bfd_of_glue_owner,
                                    tmp_name, BSF_FUNCTION | BSF_LOCAL, s, val,
                                    NULL, TRUE, FALSE, &bh);
 
  myh = (struct elf_link_hash_entry *) bh;
  myh->type = ELF_ST_INFO (STB_LOCAL, STT_FUNC);
  myh->forced_local = 1;
 
  /* Link veneer back to calling location.  */
  sec_data->erratumcount += 1;
  newerr = (elf32_vfp11_erratum_list *)
      bfd_zmalloc (sizeof (elf32_vfp11_erratum_list));
 
  newerr->type = VFP11_ERRATUM_ARM_VENEER;
  newerr->vma = -1;
  newerr->u.v.branch = branch;
  newerr->u.v.id = hash_table->num_vfp11_fixes;
  branch->u.b.veneer = newerr;
 
  newerr->next = sec_data->erratumlist;
  sec_data->erratumlist = newerr;
 
  /* A symbol for the return from the veneer.  */
  sprintf (tmp_name, VFP11_ERRATUM_VENEER_ENTRY_NAME "_r",
	   hash_table->num_vfp11_fixes);
 
  myh = elf_link_hash_lookup
    (&(hash_table)->root, tmp_name, FALSE, FALSE, FALSE);
 
  if (myh != NULL)
    abort ();
 
  bh = NULL;
  val = offset + 4;
  _bfd_generic_link_add_one_symbol (link_info, branch_bfd, tmp_name, BSF_LOCAL,
				    branch_sec, val, NULL, TRUE, FALSE, &bh);
 
  myh = (struct elf_link_hash_entry *) bh;
  myh->type = ELF_ST_INFO (STB_LOCAL, STT_FUNC);
  myh->forced_local = 1;
 
  free (tmp_name);
 
  /* Generate a mapping symbol for the veneer section, and explicitly add an
     entry for that symbol to the code/data map for the section.  */
  if (hash_table->vfp11_erratum_glue_size == 0)
    {
      bh = NULL;
      /* FIXME: Creates an ARM symbol.  Thumb mode will need attention if it
         ever requires this erratum fix.  */
      _bfd_generic_link_add_one_symbol (link_info,
					hash_table->bfd_of_glue_owner, "$a",
					BSF_LOCAL, s, 0, NULL,
                                        TRUE, FALSE, &bh);
 
      myh = (struct elf_link_hash_entry *) bh;
      myh->type = ELF_ST_INFO (STB_LOCAL, STT_NOTYPE);
      myh->forced_local = 1;
 
      /* The elf32_arm_init_maps function only cares about symbols from input
         BFDs.  We must make a note of this generated mapping symbol
         ourselves so that code byteswapping works properly in
         elf32_arm_write_section.  */
      elf32_arm_section_map_add (s, 'a', 0);
    }
 
  s->size += VFP11_ERRATUM_VENEER_SIZE;
  hash_table->vfp11_erratum_glue_size += VFP11_ERRATUM_VENEER_SIZE;
  hash_table->num_vfp11_fixes++;
 
  /* The offset of the veneer.  */
  return val;
}
 
#define ARM_GLUE_SECTION_FLAGS \
  (SEC_ALLOC | SEC_LOAD | SEC_HAS_CONTENTS | SEC_IN_MEMORY | SEC_CODE \
   | SEC_READONLY | SEC_LINKER_CREATED)
 
/* Create a fake section for use by the ARM backend of the linker.  */
 
static bfd_boolean
arm_make_glue_section (bfd * abfd, const char * name)
{
  asection * sec;
 
  sec = bfd_get_section_by_name (abfd, name);
  if (sec != NULL)
    /* Already made.  */
    return TRUE;
 
  sec = bfd_make_section_with_flags (abfd, name, ARM_GLUE_SECTION_FLAGS);
 
  if (sec == NULL
      || !bfd_set_section_alignment (abfd, sec, 2))
    return FALSE;
 
  /* Set the gc mark to prevent the section from being removed by garbage
     collection, despite the fact that no relocs refer to this section.  */
  sec->gc_mark = 1;
 
  return TRUE;
}
 
/* Add the glue sections to ABFD.  This function is called from the
   linker scripts in ld/emultempl/{armelf}.em.  */
 
bfd_boolean
bfd_elf32_arm_add_glue_sections_to_bfd (bfd *abfd,
					struct bfd_link_info *info)
{
  /* If we are only performing a partial
     link do not bother adding the glue.  */
  if (info->relocatable)
    return TRUE;
 
  return arm_make_glue_section (abfd, ARM2THUMB_GLUE_SECTION_NAME)
    && arm_make_glue_section (abfd, THUMB2ARM_GLUE_SECTION_NAME)
    && arm_make_glue_section (abfd, VFP11_ERRATUM_VENEER_SECTION_NAME)
    && arm_make_glue_section (abfd, ARM_BX_GLUE_SECTION_NAME);
}
 
/* Select a BFD to be used to hold the sections used by the glue code.
   This function is called from the linker scripts in ld/emultempl/
   {armelf/pe}.em.  */
 
bfd_boolean
bfd_elf32_arm_get_bfd_for_interworking (bfd *abfd, struct bfd_link_info *info)
{
  struct elf32_arm_link_hash_table *globals;
 
  /* If we are only performing a partial link
     do not bother getting a bfd to hold the glue.  */
  if (info->relocatable)
    return TRUE;
 
  /* Make sure we don't attach the glue sections to a dynamic object.  */
  BFD_ASSERT (!(abfd->flags & DYNAMIC));
 
  globals = elf32_arm_hash_table (info);
  BFD_ASSERT (globals != NULL);
 
  if (globals->bfd_of_glue_owner != NULL)
    return TRUE;
 
  /* Save the bfd for later use.  */
  globals->bfd_of_glue_owner = abfd;
 
  return TRUE;
}
 
static void
check_use_blx (struct elf32_arm_link_hash_table *globals)
{
  if (bfd_elf_get_obj_attr_int (globals->obfd, OBJ_ATTR_PROC,
				Tag_CPU_arch) > 2)
    globals->use_blx = 1;
}
 
bfd_boolean
bfd_elf32_arm_process_before_allocation (bfd *abfd,
					 struct bfd_link_info *link_info)
{
  Elf_Internal_Shdr *symtab_hdr;
  Elf_Internal_Rela *internal_relocs = NULL;
  Elf_Internal_Rela *irel, *irelend;
  bfd_byte *contents = NULL;
 
  asection *sec;
  struct elf32_arm_link_hash_table *globals;
 
  /* If we are only performing a partial link do not bother
     to construct any glue.  */
  if (link_info->relocatable)
    return TRUE;
 
  /* Here we have a bfd that is to be included on the link.  We have a
     hook to do reloc rummaging, before section sizes are nailed down.  */
  globals = elf32_arm_hash_table (link_info);
  BFD_ASSERT (globals != NULL);
 
  check_use_blx (globals);
 
  if (globals->byteswap_code && !bfd_big_endian (abfd))
    {
      _bfd_error_handler (_("%B: BE8 images only valid in big-endian mode."),
			  abfd);
      return FALSE;
    }
 
  /* PR 5398: If we have not decided to include any loadable sections in
     the output then we will not have a glue owner bfd.  This is OK, it
     just means that there is nothing else for us to do here.  */
  if (globals->bfd_of_glue_owner == NULL)
    return TRUE;
 
  /* Rummage around all the relocs and map the glue vectors.  */
  sec = abfd->sections;
 
  if (sec == NULL)
    return TRUE;
 
  for (; sec != NULL; sec = sec->next)
    {
      if (sec->reloc_count == 0)
	continue;
 
      if ((sec->flags & SEC_EXCLUDE) != 0)
	continue;
 
      symtab_hdr = & elf_symtab_hdr (abfd);
 
      /* Load the relocs.  */
      internal_relocs
	= _bfd_elf_link_read_relocs (abfd, sec, NULL, NULL, FALSE);
 
      if (internal_relocs == NULL)
	goto error_return;
 
      irelend = internal_relocs + sec->reloc_count;
      for (irel = internal_relocs; irel < irelend; irel++)
	{
	  long r_type;
	  unsigned long r_index;
 
	  struct elf_link_hash_entry *h;
 
	  r_type = ELF32_R_TYPE (irel->r_info);
	  r_index = ELF32_R_SYM (irel->r_info);
 
	  /* These are the only relocation types we care about.  */
	  if (   r_type != R_ARM_PC24
	      && (r_type != R_ARM_V4BX || globals->fix_v4bx < 2))
	    continue;
 
	  /* Get the section contents if we haven't done so already.  */
	  if (contents == NULL)
	    {
	      /* Get cached copy if it exists.  */
	      if (elf_section_data (sec)->this_hdr.contents != NULL)
		contents = elf_section_data (sec)->this_hdr.contents;
	      else
		{
		  /* Go get them off disk.  */
		  if (! bfd_malloc_and_get_section (abfd, sec, &contents))
		    goto error_return;
		}
	    }
 
	  if (r_type == R_ARM_V4BX)
	    {
	      int reg;
 
	      reg = bfd_get_32 (abfd, contents + irel->r_offset) & 0xf;
	      record_arm_bx_glue (link_info, reg);
	      continue;
	    }
 
	  /* If the relocation is not against a symbol it cannot concern us.  */
	  h = NULL;
 
	  /* We don't care about local symbols.  */
	  if (r_index < symtab_hdr->sh_info)
	    continue;
 
	  /* This is an external symbol.  */
	  r_index -= symtab_hdr->sh_info;
	  h = (struct elf_link_hash_entry *)
	    elf_sym_hashes (abfd)[r_index];
 
	  /* If the relocation is against a static symbol it must be within
	     the current section and so cannot be a cross ARM/Thumb relocation.  */
	  if (h == NULL)
	    continue;
 
	  /* If the call will go through a PLT entry then we do not need
	     glue.  */
	  if (globals->root.splt != NULL && h->plt.offset != (bfd_vma) -1)
	    continue;
 
	  switch (r_type)
	    {
	    case R_ARM_PC24:
	      /* This one is a call from arm code.  We need to look up
	         the target of the call.  If it is a thumb target, we
	         insert glue.  */
	      if (h->target_internal == ST_BRANCH_TO_THUMB)
		record_arm_to_thumb_glue (link_info, h);
	      break;
 
	    default:
	      abort ();
	    }
	}
 
      if (contents != NULL
	  && elf_section_data (sec)->this_hdr.contents != contents)
	free (contents);
      contents = NULL;
 
      if (internal_relocs != NULL
	  && elf_section_data (sec)->relocs != internal_relocs)
	free (internal_relocs);
      internal_relocs = NULL;
    }
 
  return TRUE;
 
error_return:
  if (contents != NULL
      && elf_section_data (sec)->this_hdr.contents != contents)
    free (contents);
  if (internal_relocs != NULL
      && elf_section_data (sec)->relocs != internal_relocs)
    free (internal_relocs);
 
  return FALSE;
}
#endif
 
 
/* Initialise maps of ARM/Thumb/data for input BFDs.  */
 
void
bfd_elf32_arm_init_maps (bfd *abfd)
{
  Elf_Internal_Sym *isymbuf;
  Elf_Internal_Shdr *hdr;
  unsigned int i, localsyms;
 
  /* PR 7093: Make sure that we are dealing with an arm elf binary.  */
  if (! is_arm_elf (abfd))
    return;
 
  if ((abfd->flags & DYNAMIC) != 0)
    return;
 
  hdr = & elf_symtab_hdr (abfd);
  localsyms = hdr->sh_info;
 
  /* Obtain a buffer full of symbols for this BFD. The hdr->sh_info field
     should contain the number of local symbols, which should come before any
     global symbols.  Mapping symbols are always local.  */
  isymbuf = bfd_elf_get_elf_syms (abfd, hdr, localsyms, 0, NULL, NULL,
				  NULL);
 
  /* No internal symbols read?  Skip this BFD.  */
  if (isymbuf == NULL)
    return;
 
  for (i = 0; i < localsyms; i++)
    {
      Elf_Internal_Sym *isym = &isymbuf[i];
      asection *sec = bfd_section_from_elf_index (abfd, isym->st_shndx);
      const char *name;
 
      if (sec != NULL
          && ELF_ST_BIND (isym->st_info) == STB_LOCAL)
        {
          name = bfd_elf_string_from_elf_section (abfd,
            hdr->sh_link, isym->st_name);
 
          if (bfd_is_arm_special_symbol_name (name,
					      BFD_ARM_SPECIAL_SYM_TYPE_MAP))
            elf32_arm_section_map_add (sec, name[1], isym->st_value);
        }
    }
}
 
 
/* Auto-select enabling of Cortex-A8 erratum fix if the user didn't explicitly
   say what they wanted.  */
 
void
bfd_elf32_arm_set_cortex_a8_fix (bfd *obfd, struct bfd_link_info *link_info)
{
  struct elf32_arm_link_hash_table *globals = elf32_arm_hash_table (link_info);
  obj_attribute *out_attr = elf_known_obj_attributes_proc (obfd);
 
  if (globals == NULL)
    return;
 
  if (globals->fix_cortex_a8 == -1)
    {
      /* Turn on Cortex-A8 erratum workaround for ARMv7-A.  */
      if (out_attr[Tag_CPU_arch].i == TAG_CPU_ARCH_V7
	  && (out_attr[Tag_CPU_arch_profile].i == 'A'
	      || out_attr[Tag_CPU_arch_profile].i == 0))
	globals->fix_cortex_a8 = 1;
      else
	globals->fix_cortex_a8 = 0;
    }
}
 
 
void
bfd_elf32_arm_set_vfp11_fix (bfd *obfd, struct bfd_link_info *link_info)
{
  struct elf32_arm_link_hash_table *globals = elf32_arm_hash_table (link_info);
  obj_attribute *out_attr = elf_known_obj_attributes_proc (obfd);
 
  if (globals == NULL)
    return;
  /* We assume that ARMv7+ does not need the VFP11 denorm erratum fix.  */
  if (out_attr[Tag_CPU_arch].i >= TAG_CPU_ARCH_V7)
    {
      switch (globals->vfp11_fix)
        {
        case BFD_ARM_VFP11_FIX_DEFAULT:
        case BFD_ARM_VFP11_FIX_NONE:
          globals->vfp11_fix = BFD_ARM_VFP11_FIX_NONE;
          break;
 
        default:
          /* Give a warning, but do as the user requests anyway.  */
          (*_bfd_error_handler) (_("%B: warning: selected VFP11 erratum "
            "workaround is not necessary for target architecture"), obfd);
        }
    }
  else if (globals->vfp11_fix == BFD_ARM_VFP11_FIX_DEFAULT)
    /* For earlier architectures, we might need the workaround, but do not
       enable it by default.  If users is running with broken hardware, they
       must enable the erratum fix explicitly.  */
    globals->vfp11_fix = BFD_ARM_VFP11_FIX_NONE;
}
 
 
enum bfd_arm_vfp11_pipe
{
  VFP11_FMAC,
  VFP11_LS,
  VFP11_DS,
  VFP11_BAD
};
 
/* Return a VFP register number.  This is encoded as RX:X for single-precision
   registers, or X:RX for double-precision registers, where RX is the group of
   four bits in the instruction encoding and X is the single extension bit.
   RX and X fields are specified using their lowest (starting) bit.  The return
   value is:
 
     0...31: single-precision registers s0...s31
     32...63: double-precision registers d0...d31.
 
   Although X should be zero for VFP11 (encoding d0...d15 only), we might
   encounter VFP3 instructions, so we allow the full range for DP registers.  */
 
static unsigned int
bfd_arm_vfp11_regno (unsigned int insn, bfd_boolean is_double, unsigned int rx,
                     unsigned int x)
{
  if (is_double)
    return (((insn >> rx) & 0xf) | (((insn >> x) & 1) << 4)) + 32;
  else
    return (((insn >> rx) & 0xf) << 1) | ((insn >> x) & 1);
}
 
/* Set bits in *WMASK according to a register number REG as encoded by
   bfd_arm_vfp11_regno().  Ignore d16-d31.  */
 
static void
bfd_arm_vfp11_write_mask (unsigned int *wmask, unsigned int reg)
{
  if (reg < 32)
    *wmask |= 1 << reg;
  else if (reg < 48)
    *wmask |= 3 << ((reg - 32) * 2);
}
 
/* Return TRUE if WMASK overwrites anything in REGS.  */
 
static bfd_boolean
bfd_arm_vfp11_antidependency (unsigned int wmask, int *regs, int numregs)
{
  int i;
 
  for (i = 0; i < numregs; i++)
    {
      unsigned int reg = regs[i];
 
      if (reg < 32 && (wmask & (1 << reg)) != 0)
        return TRUE;
 
      reg -= 32;
 
      if (reg >= 16)
        continue;
 
      if ((wmask & (3 << (reg * 2))) != 0)
        return TRUE;
    }
 
  return FALSE;
}
 
/* In this function, we're interested in two things: finding input registers
   for VFP data-processing instructions, and finding the set of registers which
   arbitrary VFP instructions may write to.  We use a 32-bit unsigned int to
   hold the written set, so FLDM etc. are easy to deal with (we're only
   interested in 32 SP registers or 16 dp registers, due to the VFP version
   implemented by the chip in question).  DP registers are marked by setting
   both SP registers in the write mask).  */
 
static enum bfd_arm_vfp11_pipe
bfd_arm_vfp11_insn_decode (unsigned int insn, unsigned int *destmask, int *regs,
                           int *numregs)
{
  enum bfd_arm_vfp11_pipe vpipe = VFP11_BAD;
  bfd_boolean is_double = ((insn & 0xf00) == 0xb00) ? 1 : 0;
 
  if ((insn & 0x0f000e10) == 0x0e000a00)  /* A data-processing insn.  */
    {
      unsigned int pqrs;
      unsigned int fd = bfd_arm_vfp11_regno (insn, is_double, 12, 22);
      unsigned int fm = bfd_arm_vfp11_regno (insn, is_double, 0, 5);
 
      pqrs = ((insn & 0x00800000) >> 20)
           | ((insn & 0x00300000) >> 19)
           | ((insn & 0x00000040) >> 6);
 
      switch (pqrs)
        {
        case 0: /* fmac[sd].  */
        case 1: /* fnmac[sd].  */
        case 2: /* fmsc[sd].  */
        case 3: /* fnmsc[sd].  */
          vpipe = VFP11_FMAC;
          bfd_arm_vfp11_write_mask (destmask, fd);
          regs[0] = fd;
          regs[1] = bfd_arm_vfp11_regno (insn, is_double, 16, 7);  /* Fn.  */
          regs[2] = fm;
          *numregs = 3;
          break;
 
        case 4: /* fmul[sd].  */
        case 5: /* fnmul[sd].  */
        case 6: /* fadd[sd].  */
        case 7: /* fsub[sd].  */
          vpipe = VFP11_FMAC;
          goto vfp_binop;
 
        case 8: /* fdiv[sd].  */
          vpipe = VFP11_DS;
          vfp_binop:
          bfd_arm_vfp11_write_mask (destmask, fd);
          regs[0] = bfd_arm_vfp11_regno (insn, is_double, 16, 7);   /* Fn.  */
          regs[1] = fm;
          *numregs = 2;
          break;
 
        case 15: /* extended opcode.  */
          {
            unsigned int extn = ((insn >> 15) & 0x1e)
                              | ((insn >> 7) & 1);
 
            switch (extn)
              {
              case 0: /* fcpy[sd].  */
              case 1: /* fabs[sd].  */
              case 2: /* fneg[sd].  */
              case 8: /* fcmp[sd].  */
              case 9: /* fcmpe[sd].  */
              case 10: /* fcmpz[sd].  */
              case 11: /* fcmpez[sd].  */
              case 16: /* fuito[sd].  */
              case 17: /* fsito[sd].  */
              case 24: /* ftoui[sd].  */
              case 25: /* ftouiz[sd].  */
              case 26: /* ftosi[sd].  */
              case 27: /* ftosiz[sd].  */
                /* These instructions will not bounce due to underflow.  */
                *numregs = 0;
                vpipe = VFP11_FMAC;
                break;
 
              case 3: /* fsqrt[sd].  */
                /* fsqrt cannot underflow, but it can (perhaps) overwrite
                   registers to cause the erratum in previous instructions.  */
                bfd_arm_vfp11_write_mask (destmask, fd);
                vpipe = VFP11_DS;
                break;
 
              case 15: /* fcvt{ds,sd}.  */
                {
                  int rnum = 0;
 
                  bfd_arm_vfp11_write_mask (destmask, fd);
 
		  /* Only FCVTSD can underflow.  */
                  if ((insn & 0x100) != 0)
                    regs[rnum++] = fm;
 
                  *numregs = rnum;
 
                  vpipe = VFP11_FMAC;
                }
                break;
 
              default:
                return VFP11_BAD;
              }
          }
          break;
 
        default:
          return VFP11_BAD;
        }
    }
  /* Two-register transfer.  */
  else if ((insn & 0x0fe00ed0) == 0x0c400a10)
    {
      unsigned int fm = bfd_arm_vfp11_regno (insn, is_double, 0, 5);
 
      if ((insn & 0x100000) == 0)
	{
          if (is_double)
            bfd_arm_vfp11_write_mask (destmask, fm);
          else
            {
              bfd_arm_vfp11_write_mask (destmask, fm);
              bfd_arm_vfp11_write_mask (destmask, fm + 1);
            }
	}
 
      vpipe = VFP11_LS;
    }
  else if ((insn & 0x0e100e00) == 0x0c100a00)  /* A load insn.  */
    {
      int fd = bfd_arm_vfp11_regno (insn, is_double, 12, 22);
      unsigned int puw = ((insn >> 21) & 0x1) | (((insn >> 23) & 3) << 1);
 
      switch (puw)
        {
        case 0: /* Two-reg transfer.  We should catch these above.  */
          abort ();
 
        case 2: /* fldm[sdx].  */
        case 3:
        case 5:
          {
            unsigned int i, offset = insn & 0xff;
 
            if (is_double)
              offset >>= 1;
 
            for (i = fd; i < fd + offset; i++)
              bfd_arm_vfp11_write_mask (destmask, i);
          }
          break;
 
        case 4: /* fld[sd].  */
        case 6:
          bfd_arm_vfp11_write_mask (destmask, fd);
          break;
 
        default:
          return VFP11_BAD;
        }
 
      vpipe = VFP11_LS;
    }
  /* Single-register transfer. Note L==0.  */
  else if ((insn & 0x0f100e10) == 0x0e000a10)
    {
      unsigned int opcode = (insn >> 21) & 7;
      unsigned int fn = bfd_arm_vfp11_regno (insn, is_double, 16, 7);
 
      switch (opcode)
        {
        case 0: /* fmsr/fmdlr.  */
        case 1: /* fmdhr.  */
          /* Mark fmdhr and fmdlr as writing to the whole of the DP
             destination register.  I don't know if this is exactly right,
             but it is the conservative choice.  */
          bfd_arm_vfp11_write_mask (destmask, fn);
          break;
 
        case 7: /* fmxr.  */
          break;
        }
 
      vpipe = VFP11_LS;
    }
 
  return vpipe;
}
 
 
static int elf32_arm_compare_mapping (const void * a, const void * b);
 
 
/* Look for potentially-troublesome code sequences which might trigger the
   VFP11 denormal/antidependency erratum.  See, e.g., the ARM1136 errata sheet
   (available from ARM) for details of the erratum.  A short version is
   described in ld.texinfo.  */
 
bfd_boolean
bfd_elf32_arm_vfp11_erratum_scan (bfd *abfd, struct bfd_link_info *link_info)
{
  asection *sec;
  bfd_byte *contents = NULL;
  int state = 0;
  int regs[3], numregs = 0;
  struct elf32_arm_link_hash_table *globals = elf32_arm_hash_table (link_info);
  int use_vector = (globals->vfp11_fix == BFD_ARM_VFP11_FIX_VECTOR);
 
  if (globals == NULL)
    return FALSE;
 
  /* We use a simple FSM to match troublesome VFP11 instruction sequences.
     The states transition as follows:
 
       0 -> 1 (vector) or 0 -> 2 (scalar)
           A VFP FMAC-pipeline instruction has been seen. Fill
           regs[0]..regs[numregs-1] with its input operands. Remember this
           instruction in 'first_fmac'.
 
       1 -> 2
           Any instruction, except for a VFP instruction which overwrites
           regs[*].
 
       1 -> 3 [ -> 0 ]  or
       2 -> 3 [ -> 0 ]
           A VFP instruction has been seen which overwrites any of regs[*].
           We must make a veneer!  Reset state to 0 before examining next
           instruction.
 
       2 -> 0
           If we fail to match anything in state 2, reset to state 0 and reset
           the instruction pointer to the instruction after 'first_fmac'.
 
     If the VFP11 vector mode is in use, there must be at least two unrelated
     instructions between anti-dependent VFP11 instructions to properly avoid
     triggering the erratum, hence the use of the extra state 1.  */
 
  /* If we are only performing a partial link do not bother
     to construct any glue.  */
  if (link_info->relocatable)
    return TRUE;
 
  /* Skip if this bfd does not correspond to an ELF image.  */
  if (! is_arm_elf (abfd))
    return TRUE;
 
  /* We should have chosen a fix type by the time we get here.  */
  BFD_ASSERT (globals->vfp11_fix != BFD_ARM_VFP11_FIX_DEFAULT);
 
  if (globals->vfp11_fix == BFD_ARM_VFP11_FIX_NONE)
    return TRUE;
 
  /* Skip this BFD if it corresponds to an executable or dynamic object.  */
  if ((abfd->flags & (EXEC_P | DYNAMIC)) != 0)
    return TRUE;
 
  for (sec = abfd->sections; sec != NULL; sec = sec->next)
    {
      unsigned int i, span, first_fmac = 0, veneer_of_insn = 0;
      struct _arm_elf_section_data *sec_data;
 
      /* If we don't have executable progbits, we're not interested in this
         section.  Also skip if section is to be excluded.  */
      if (elf_section_type (sec) != SHT_PROGBITS
          || (elf_section_flags (sec) & SHF_EXECINSTR) == 0
          || (sec->flags & SEC_EXCLUDE) != 0
	  || sec->sec_info_type == ELF_INFO_TYPE_JUST_SYMS
	  || sec->output_section == bfd_abs_section_ptr
          || strcmp (sec->name, VFP11_ERRATUM_VENEER_SECTION_NAME) == 0)
        continue;
 
      sec_data = elf32_arm_section_data (sec);
 
      if (sec_data->mapcount == 0)
        continue;
 
      if (elf_section_data (sec)->this_hdr.contents != NULL)
	contents = elf_section_data (sec)->this_hdr.contents;
      else if (! bfd_malloc_and_get_section (abfd, sec, &contents))
	goto error_return;
 
      qsort (sec_data->map, sec_data->mapcount, sizeof (elf32_arm_section_map),
	     elf32_arm_compare_mapping);
 
      for (span = 0; span < sec_data->mapcount; span++)
        {
          unsigned int span_start = sec_data->map[span].vma;
          unsigned int span_end = (span == sec_data->mapcount - 1)
				  ? sec->size : sec_data->map[span + 1].vma;
          char span_type = sec_data->map[span].type;
 
          /* FIXME: Only ARM mode is supported at present.  We may need to
             support Thumb-2 mode also at some point.  */
          if (span_type != 'a')
            continue;
 
          for (i = span_start; i < span_end;)
            {
              unsigned int next_i = i + 4;
              unsigned int insn = bfd_big_endian (abfd)
                ? (contents[i] << 24)
                  | (contents[i + 1] << 16)
                  | (contents[i + 2] << 8)
                  | contents[i + 3]
                : (contents[i + 3] << 24)
                  | (contents[i + 2] << 16)
                  | (contents[i + 1] << 8)
                  | contents[i];
              unsigned int writemask = 0;
              enum bfd_arm_vfp11_pipe vpipe;
 
              switch (state)
                {
                case 0:
                  vpipe = bfd_arm_vfp11_insn_decode (insn, &writemask, regs,
                                                    &numregs);
                  /* I'm assuming the VFP11 erratum can trigger with denorm
                     operands on either the FMAC or the DS pipeline. This might
                     lead to slightly overenthusiastic veneer insertion.  */
                  if (vpipe == VFP11_FMAC || vpipe == VFP11_DS)
                    {
                      state = use_vector ? 1 : 2;
                      first_fmac = i;
                      veneer_of_insn = insn;
                    }
                  break;
 
                case 1:
                  {
                    int other_regs[3], other_numregs;
                    vpipe = bfd_arm_vfp11_insn_decode (insn, &writemask,
						      other_regs,
                                                      &other_numregs);
                    if (vpipe != VFP11_BAD
                        && bfd_arm_vfp11_antidependency (writemask, regs,
							 numregs))
                      state = 3;
                    else
                      state = 2;
                  }
                  break;
 
                case 2:
                  {
                    int other_regs[3], other_numregs;
                    vpipe = bfd_arm_vfp11_insn_decode (insn, &writemask,
						      other_regs,
                                                      &other_numregs);
                    if (vpipe != VFP11_BAD
                        && bfd_arm_vfp11_antidependency (writemask, regs,
							 numregs))
                      state = 3;
                    else
                      {
                        state = 0;
                        next_i = first_fmac + 4;
                      }
                  }
                  break;
 
                case 3:
                  abort ();  /* Should be unreachable.  */
                }
 
              if (state == 3)
                {
                  elf32_vfp11_erratum_list *newerr =(elf32_vfp11_erratum_list *)
                      bfd_zmalloc (sizeof (elf32_vfp11_erratum_list));
 
                  elf32_arm_section_data (sec)->erratumcount += 1;
 
                  newerr->u.b.vfp_insn = veneer_of_insn;
 
                  switch (span_type)
                    {
                    case 'a':
                      newerr->type = VFP11_ERRATUM_BRANCH_TO_ARM_VENEER;
                      break;
 
                    default:
                      abort ();
                    }
 
                  record_vfp11_erratum_veneer (link_info, newerr, abfd, sec,
					       first_fmac);
 
                  newerr->vma = -1;
 
                  newerr->next = sec_data->erratumlist;
                  sec_data->erratumlist = newerr;
 
                  state = 0;
                }
 
              i = next_i;
            }
        }
 
      if (contents != NULL
          && elf_section_data (sec)->this_hdr.contents != contents)
        free (contents);
      contents = NULL;
    }
 
  return TRUE;
 
error_return:
  if (contents != NULL
      && elf_section_data (sec)->this_hdr.contents != contents)
    free (contents);
 
  return FALSE;
}
 
/* Find virtual-memory addresses for VFP11 erratum veneers and return locations
   after sections have been laid out, using specially-named symbols.  */
 
void
bfd_elf32_arm_vfp11_fix_veneer_locations (bfd *abfd,
					  struct bfd_link_info *link_info)
{
  asection *sec;
  struct elf32_arm_link_hash_table *globals;
  char *tmp_name;
 
  if (link_info->relocatable)
    return;
 
  /* Skip if this bfd does not correspond to an ELF image.  */
  if (! is_arm_elf (abfd))
    return;
 
  globals = elf32_arm_hash_table (link_info);
  if (globals == NULL)
    return;
 
  tmp_name = (char *) bfd_malloc ((bfd_size_type) strlen
                                  (VFP11_ERRATUM_VENEER_ENTRY_NAME) + 10);
 
  for (sec = abfd->sections; sec != NULL; sec = sec->next)
    {
      struct _arm_elf_section_data *sec_data = elf32_arm_section_data (sec);
      elf32_vfp11_erratum_list *errnode = sec_data->erratumlist;
 
      for (; errnode != NULL; errnode = errnode->next)
        {
          struct elf_link_hash_entry *myh;
          bfd_vma vma;
 
          switch (errnode->type)
            {
            case VFP11_ERRATUM_BRANCH_TO_ARM_VENEER:
            case VFP11_ERRATUM_BRANCH_TO_THUMB_VENEER:
              /* Find veneer symbol.  */
              sprintf (tmp_name, VFP11_ERRATUM_VENEER_ENTRY_NAME,
		       errnode->u.b.veneer->u.v.id);
 
              myh = elf_link_hash_lookup
                (&(globals)->root, tmp_name, FALSE, FALSE, TRUE);
 
              if (myh == NULL)
                (*_bfd_error_handler) (_("%B: unable to find VFP11 veneer "
                			 "`%s'"), abfd, tmp_name);
 
              vma = myh->root.u.def.section->output_section->vma
                    + myh->root.u.def.section->output_offset
                    + myh->root.u.def.value;
 
              errnode->u.b.veneer->vma = vma;
              break;
 
	    case VFP11_ERRATUM_ARM_VENEER:
            case VFP11_ERRATUM_THUMB_VENEER:
              /* Find return location.  */
              sprintf (tmp_name, VFP11_ERRATUM_VENEER_ENTRY_NAME "_r",
                       errnode->u.v.id);
 
              myh = elf_link_hash_lookup
                (&(globals)->root, tmp_name, FALSE, FALSE, TRUE);
 
              if (myh == NULL)
                (*_bfd_error_handler) (_("%B: unable to find VFP11 veneer "
					 "`%s'"), abfd, tmp_name);
 
              vma = myh->root.u.def.section->output_section->vma
                    + myh->root.u.def.section->output_offset
                    + myh->root.u.def.value;
 
              errnode->u.v.branch->vma = vma;
              break;
 
            default:
              abort ();
            }
        }
    }
 
  free (tmp_name);
}
 
 
/* Set target relocation values needed during linking.  */
 
void
bfd_elf32_arm_set_target_relocs (struct bfd *output_bfd,
				 struct bfd_link_info *link_info,
				 int target1_is_rel,
				 char * target2_type,
                                 int fix_v4bx,
				 int use_blx,
                                 bfd_arm_vfp11_fix vfp11_fix,
				 int no_enum_warn, int no_wchar_warn,
				 int pic_veneer, int fix_cortex_a8)
{
  struct elf32_arm_link_hash_table *globals;
 
  globals = elf32_arm_hash_table (link_info);
  if (globals == NULL)
    return;
 
  globals->target1_is_rel = target1_is_rel;
  if (strcmp (target2_type, "rel") == 0)
    globals->target2_reloc = R_ARM_REL32;
  else if (strcmp (target2_type, "abs") == 0)
    globals->target2_reloc = R_ARM_ABS32;
  else if (strcmp (target2_type, "got-rel") == 0)
    globals->target2_reloc = R_ARM_GOT_PREL;
  else
    {
      _bfd_error_handler (_("Invalid TARGET2 relocation type '%s'."),
			  target2_type);
    }
  globals->fix_v4bx = fix_v4bx;
  globals->use_blx |= use_blx;
  globals->vfp11_fix = vfp11_fix;
  globals->pic_veneer = pic_veneer;
  globals->fix_cortex_a8 = fix_cortex_a8;
 
  BFD_ASSERT (is_arm_elf (output_bfd));
  elf_arm_tdata (output_bfd)->no_enum_size_warning = no_enum_warn;
  elf_arm_tdata (output_bfd)->no_wchar_size_warning = no_wchar_warn;
}
 
/* Replace the target offset of a Thumb bl or b.w instruction.  */
 
static void
insert_thumb_branch (bfd *abfd, long int offset, bfd_byte *insn)
{
  bfd_vma upper;
  bfd_vma lower;
  int reloc_sign;
 
  BFD_ASSERT ((offset & 1) == 0);
 
  upper = bfd_get_16 (abfd, insn);
  lower = bfd_get_16 (abfd, insn + 2);
  reloc_sign = (offset < 0) ? 1 : 0;
  upper = (upper & ~(bfd_vma) 0x7ff)
	  | ((offset >> 12) & 0x3ff)
	  | (reloc_sign << 10);
  lower = (lower & ~(bfd_vma) 0x2fff)
	  | (((!((offset >> 23) & 1)) ^ reloc_sign) << 13)
	  | (((!((offset >> 22) & 1)) ^ reloc_sign) << 11)
	  | ((offset >> 1) & 0x7ff);
  bfd_put_16 (abfd, upper, insn);
  bfd_put_16 (abfd, lower, insn + 2);
}
 
/* Thumb code calling an ARM function.  */
 
static int
elf32_thumb_to_arm_stub (struct bfd_link_info * info,
			 const char *           name,
			 bfd *                  input_bfd,
			 bfd *                  output_bfd,
			 asection *             input_section,
			 bfd_byte *             hit_data,
			 asection *             sym_sec,
			 bfd_vma                offset,
			 bfd_signed_vma         addend,
			 bfd_vma                val,
			 char **error_message)
{
  asection * s = 0;
  bfd_vma my_offset;
  long int ret_offset;
  struct elf_link_hash_entry * myh;
  struct elf32_arm_link_hash_table * globals;
 
  myh = find_thumb_glue (info, name, error_message);
  if (myh == NULL)
    return FALSE;
 
  globals = elf32_arm_hash_table (info);
  BFD_ASSERT (globals != NULL);
  BFD_ASSERT (globals->bfd_of_glue_owner != NULL);
 
  my_offset = myh->root.u.def.value;
 
  s = bfd_get_section_by_name (globals->bfd_of_glue_owner,
			       THUMB2ARM_GLUE_SECTION_NAME);
 
  BFD_ASSERT (s != NULL);
  BFD_ASSERT (s->contents != NULL);
  BFD_ASSERT (s->output_section != NULL);
 
  if ((my_offset & 0x01) == 0x01)
    {
      if (sym_sec != NULL
	  && sym_sec->owner != NULL
	  && !INTERWORK_FLAG (sym_sec->owner))
	{
	  (*_bfd_error_handler)
	    (_("%B(%s): warning: interworking not enabled.\n"
	       "  first occurrence: %B: thumb call to arm"),
	     sym_sec->owner, input_bfd, name);
 
	  return FALSE;
	}
 
      --my_offset;
      myh->root.u.def.value = my_offset;
 
      put_thumb_insn (globals, output_bfd, (bfd_vma) t2a1_bx_pc_insn,
		      s->contents + my_offset);
 
      put_thumb_insn (globals, output_bfd, (bfd_vma) t2a2_noop_insn,
		      s->contents + my_offset + 2);
 
      ret_offset =
	/* Address of destination of the stub.  */
	((bfd_signed_vma) val)
	- ((bfd_signed_vma)
	   /* Offset from the start of the current section
	      to the start of the stubs.  */
	   (s->output_offset
	    /* Offset of the start of this stub from the start of the stubs.  */
	    + my_offset
	    /* Address of the start of the current section.  */
	    + s->output_section->vma)
	   /* The branch instruction is 4 bytes into the stub.  */
	   + 4
	   /* ARM branches work from the pc of the instruction + 8.  */
	   + 8);
 
      put_arm_insn (globals, output_bfd,
		    (bfd_vma) t2a3_b_insn | ((ret_offset >> 2) & 0x00FFFFFF),
		    s->contents + my_offset + 4);
    }
 
  BFD_ASSERT (my_offset <= globals->thumb_glue_size);
 
  /* Now go back and fix up the original BL insn to point to here.  */
  ret_offset =
    /* Address of where the stub is located.  */
    (s->output_section->vma + s->output_offset + my_offset)
     /* Address of where the BL is located.  */
    - (input_section->output_section->vma + input_section->output_offset
       + offset)
    /* Addend in the relocation.  */
    - addend
    /* Biassing for PC-relative addressing.  */
    - 8;
 
  insert_thumb_branch (input_bfd, ret_offset, hit_data - input_section->vma);
 
  return TRUE;
}
 
/* Populate an Arm to Thumb stub.  Returns the stub symbol.  */
 
static struct elf_link_hash_entry *
elf32_arm_create_thumb_stub (struct bfd_link_info * info,
			     const char *           name,
			     bfd *                  input_bfd,
			     bfd *                  output_bfd,
			     asection *             sym_sec,
			     bfd_vma                val,
			     asection *             s,
			     char **                error_message)
{
  bfd_vma my_offset;
  long int ret_offset;
  struct elf_link_hash_entry * myh;
  struct elf32_arm_link_hash_table * globals;
 
  myh = find_arm_glue (info, name, error_message);
  if (myh == NULL)
    return NULL;
 
  globals = elf32_arm_hash_table (info);
  BFD_ASSERT (globals != NULL);
  BFD_ASSERT (globals->bfd_of_glue_owner != NULL);
 
  my_offset = myh->root.u.def.value;
 
  if ((my_offset & 0x01) == 0x01)
    {
      if (sym_sec != NULL
	  && sym_sec->owner != NULL
	  && !INTERWORK_FLAG (sym_sec->owner))
	{
	  (*_bfd_error_handler)
	    (_("%B(%s): warning: interworking not enabled.\n"
	       "  first occurrence: %B: arm call to thumb"),
	     sym_sec->owner, input_bfd, name);
	}
 
      --my_offset;
      myh->root.u.def.value = my_offset;
 
      if (info->shared || globals->root.is_relocatable_executable
	  || globals->pic_veneer)
	{
	  /* For relocatable objects we can't use absolute addresses,
	     so construct the address from a relative offset.  */
	  /* TODO: If the offset is small it's probably worth
	     constructing the address with adds.  */
	  put_arm_insn (globals, output_bfd, (bfd_vma) a2t1p_ldr_insn,
			s->contents + my_offset);
	  put_arm_insn (globals, output_bfd, (bfd_vma) a2t2p_add_pc_insn,
			s->contents + my_offset + 4);
	  put_arm_insn (globals, output_bfd, (bfd_vma) a2t3p_bx_r12_insn,
			s->contents + my_offset + 8);
	  /* Adjust the offset by 4 for the position of the add,
	     and 8 for the pipeline offset.  */
	  ret_offset = (val - (s->output_offset
			       + s->output_section->vma
			       + my_offset + 12))
		       | 1;
	  bfd_put_32 (output_bfd, ret_offset,
		      s->contents + my_offset + 12);
	}
      else if (globals->use_blx)
	{
	  put_arm_insn (globals, output_bfd, (bfd_vma) a2t1v5_ldr_insn,
			s->contents + my_offset);
 
	  /* It's a thumb address.  Add the low order bit.  */
	  bfd_put_32 (output_bfd, val | a2t2v5_func_addr_insn,
		      s->contents + my_offset + 4);
	}
      else
	{
	  put_arm_insn (globals, output_bfd, (bfd_vma) a2t1_ldr_insn,
			s->contents + my_offset);
 
	  put_arm_insn (globals, output_bfd, (bfd_vma) a2t2_bx_r12_insn,
			s->contents + my_offset + 4);
 
	  /* It's a thumb address.  Add the low order bit.  */
	  bfd_put_32 (output_bfd, val | a2t3_func_addr_insn,
		      s->contents + my_offset + 8);
 
	  my_offset += 12;
	}
    }
 
  BFD_ASSERT (my_offset <= globals->arm_glue_size);
 
  return myh;
}
 
/* Arm code calling a Thumb function.  */
 
static int
elf32_arm_to_thumb_stub (struct bfd_link_info * info,
			 const char *           name,
			 bfd *                  input_bfd,
			 bfd *                  output_bfd,
			 asection *             input_section,
			 bfd_byte *             hit_data,
			 asection *             sym_sec,
			 bfd_vma                offset,
			 bfd_signed_vma         addend,
			 bfd_vma                val,
			 char **error_message)
{
  unsigned long int tmp;
  bfd_vma my_offset;
  asection * s;
  long int ret_offset;
  struct elf_link_hash_entry * myh;
  struct elf32_arm_link_hash_table * globals;
 
  globals = elf32_arm_hash_table (info);
  BFD_ASSERT (globals != NULL);
  BFD_ASSERT (globals->bfd_of_glue_owner != NULL);
 
  s = bfd_get_section_by_name (globals->bfd_of_glue_owner,
			       ARM2THUMB_GLUE_SECTION_NAME);
  BFD_ASSERT (s != NULL);
  BFD_ASSERT (s->contents != NULL);
  BFD_ASSERT (s->output_section != NULL);
 
  myh = elf32_arm_create_thumb_stub (info, name, input_bfd, output_bfd,
				     sym_sec, val, s, error_message);
  if (!myh)
    return FALSE;
 
  my_offset = myh->root.u.def.value;
  tmp = bfd_get_32 (input_bfd, hit_data);
  tmp = tmp & 0xFF000000;
 
  /* Somehow these are both 4 too far, so subtract 8.  */
  ret_offset = (s->output_offset
		+ my_offset
		+ s->output_section->vma
		- (input_section->output_offset
		   + input_section->output_section->vma
		   + offset + addend)
		- 8);
 
  tmp = tmp | ((ret_offset >> 2) & 0x00FFFFFF);
 
  bfd_put_32 (output_bfd, (bfd_vma) tmp, hit_data - input_section->vma);
 
  return TRUE;
}
 
/* Populate Arm stub for an exported Thumb function.  */
 
static bfd_boolean
elf32_arm_to_thumb_export_stub (struct elf_link_hash_entry *h, void * inf)
{
  struct bfd_link_info * info = (struct bfd_link_info *) inf;
  asection * s;
  struct elf_link_hash_entry * myh;
  struct elf32_arm_link_hash_entry *eh;
  struct elf32_arm_link_hash_table * globals;
  asection *sec;
  bfd_vma val;
  char *error_message;
 
  eh = elf32_arm_hash_entry (h);
  /* Allocate stubs for exported Thumb functions on v4t.  */
  if (eh->export_glue == NULL)
    return TRUE;
 
  globals = elf32_arm_hash_table (info);
  BFD_ASSERT (globals != NULL);
  BFD_ASSERT (globals->bfd_of_glue_owner != NULL);
 
  s = bfd_get_section_by_name (globals->bfd_of_glue_owner,
			       ARM2THUMB_GLUE_SECTION_NAME);
  BFD_ASSERT (s != NULL);
  BFD_ASSERT (s->contents != NULL);
  BFD_ASSERT (s->output_section != NULL);
 
  sec = eh->export_glue->root.u.def.section;
 
  BFD_ASSERT (sec->output_section != NULL);
 
  val = eh->export_glue->root.u.def.value + sec->output_offset
	+ sec->output_section->vma;
 
  myh = elf32_arm_create_thumb_stub (info, h->root.root.string,
				     h->root.u.def.section->owner,
				     globals->obfd, sec, val, s,
				     &error_message);
  BFD_ASSERT (myh);
  return TRUE;
}
 
/* Populate ARMv4 BX veneers.  Returns the absolute adress of the veneer.  */
 
static bfd_vma
elf32_arm_bx_glue (struct bfd_link_info * info, int reg)
{
  bfd_byte *p;
  bfd_vma glue_addr;
  asection *s;
  struct elf32_arm_link_hash_table *globals;
 
  globals = elf32_arm_hash_table (info);
  BFD_ASSERT (globals != NULL);
  BFD_ASSERT (globals->bfd_of_glue_owner != NULL);
 
  s = bfd_get_section_by_name (globals->bfd_of_glue_owner,
			       ARM_BX_GLUE_SECTION_NAME);
  BFD_ASSERT (s != NULL);
  BFD_ASSERT (s->contents != NULL);
  BFD_ASSERT (s->output_section != NULL);
 
  BFD_ASSERT (globals->bx_glue_offset[reg] & 2);
 
  glue_addr = globals->bx_glue_offset[reg] & ~(bfd_vma)3;
 
  if ((globals->bx_glue_offset[reg] & 1) == 0)
    {
      p = s->contents + glue_addr;
      bfd_put_32 (globals->obfd, armbx1_tst_insn + (reg << 16), p);
      bfd_put_32 (globals->obfd, armbx2_moveq_insn + reg, p + 4);
      bfd_put_32 (globals->obfd, armbx3_bx_insn + reg, p + 8);
      globals->bx_glue_offset[reg] |= 1;
    }
 
  return glue_addr + s->output_section->vma + s->output_offset;
}
 
/* Generate Arm stubs for exported Thumb symbols.  */
static void
elf32_arm_begin_write_processing (bfd *abfd ATTRIBUTE_UNUSED,
				  struct bfd_link_info *link_info)
{
  struct elf32_arm_link_hash_table * globals;
 
  if (link_info == NULL)
    /* Ignore this if we are not called by the ELF backend linker.  */
    return;
 
  globals = elf32_arm_hash_table (link_info);
  if (globals == NULL)
    return;
 
  /* If blx is available then exported Thumb symbols are OK and there is
     nothing to do.  */
  if (globals->use_blx)
    return;
 
  elf_link_hash_traverse (&globals->root, elf32_arm_to_thumb_export_stub,
			  link_info);
}
 
/* Reserve space for COUNT dynamic relocations in relocation selection
   SRELOC.  */
 
static void
elf32_arm_allocate_dynrelocs (struct bfd_link_info *info, asection *sreloc,
			      bfd_size_type count)
{
  struct elf32_arm_link_hash_table *htab;
 
  htab = elf32_arm_hash_table (info);
  BFD_ASSERT (htab->root.dynamic_sections_created);
  if (sreloc == NULL)
    abort ();
  sreloc->size += RELOC_SIZE (htab) * count;
}
 
/* Reserve space for COUNT R_ARM_IRELATIVE relocations.  If the link is
   dynamic, the relocations should go in SRELOC, otherwise they should
   go in the special .rel.iplt section.  */
 
static void
elf32_arm_allocate_irelocs (struct bfd_link_info *info, asection *sreloc,
			    bfd_size_type count)
{
  struct elf32_arm_link_hash_table *htab;
 
  htab = elf32_arm_hash_table (info);
  if (!htab->root.dynamic_sections_created)
    htab->root.irelplt->size += RELOC_SIZE (htab) * count;
  else
    {
      BFD_ASSERT (sreloc != NULL);
      sreloc->size += RELOC_SIZE (htab) * count;
    }
}
 
/* Add relocation REL to the end of relocation section SRELOC.  */
 
static void
elf32_arm_add_dynreloc (bfd *output_bfd, struct bfd_link_info *info,
			asection *sreloc, Elf_Internal_Rela *rel)
{
  bfd_byte *loc;
  struct elf32_arm_link_hash_table *htab;
 
  htab = elf32_arm_hash_table (info);
  if (!htab->root.dynamic_sections_created
      && ELF32_R_TYPE (rel->r_info) == R_ARM_IRELATIVE)
    sreloc = htab->root.irelplt;
  if (sreloc == NULL)
    abort ();
  loc = sreloc->contents;
  loc += sreloc->reloc_count++ * RELOC_SIZE (htab);
  if (sreloc->reloc_count * RELOC_SIZE (htab) > sreloc->size)
    abort ();
  SWAP_RELOC_OUT (htab) (output_bfd, rel, loc);
}
 
/* Allocate room for a PLT entry described by ROOT_PLT and ARM_PLT.
   IS_IPLT_ENTRY says whether the entry belongs to .iplt rather than
   to .plt.  */
 
static void
elf32_arm_allocate_plt_entry (struct bfd_link_info *info,
			      bfd_boolean is_iplt_entry,
			      union gotplt_union *root_plt,
			      struct arm_plt_info *arm_plt)
{
  struct elf32_arm_link_hash_table *htab;
  asection *splt;
  asection *sgotplt;
 
  htab = elf32_arm_hash_table (info);
 
  if (is_iplt_entry)
    {
      splt = htab->root.iplt;
      sgotplt = htab->root.igotplt;
 
      /* Allocate room for an R_ARM_IRELATIVE relocation in .rel.iplt.  */
      elf32_arm_allocate_irelocs (info, htab->root.irelplt, 1);
    }
  else
    {
      splt = htab->root.splt;
      sgotplt = htab->root.sgotplt;
 
      /* Allocate room for an R_JUMP_SLOT relocation in .rel.plt.  */
      elf32_arm_allocate_dynrelocs (info, htab->root.srelplt, 1);
 
      /* If this is the first .plt entry, make room for the special
	 first entry.  */
      if (splt->size == 0)
	splt->size += htab->plt_header_size;
    }
 
  /* Allocate the PLT entry itself, including any leading Thumb stub.  */
  if (elf32_arm_plt_needs_thumb_stub_p (info, arm_plt))
    splt->size += PLT_THUMB_STUB_SIZE;
  root_plt->offset = splt->size;
  splt->size += htab->plt_entry_size;
 
  if (!htab->symbian_p)
    {
      /* We also need to make an entry in the .got.plt section, which
	 will be placed in the .got section by the linker script.  */
      arm_plt->got_offset = sgotplt->size - 8 * htab->num_tls_desc;
      sgotplt->size += 4;
    }
}
 
/* Fill in a PLT entry and its associated GOT slot.  If DYNINDX == -1,
   the entry lives in .iplt and resolves to (*SYM_VALUE)().
   Otherwise, DYNINDX is the index of the symbol in the dynamic
   symbol table and SYM_VALUE is undefined.
 
   ROOT_PLT points to the offset of the PLT entry from the start of its
   section (.iplt or .plt).  ARM_PLT points to the symbol's ARM-specific
   bookkeeping information.  */
 
static void
elf32_arm_populate_plt_entry (bfd *output_bfd, struct bfd_link_info *info,
			      union gotplt_union *root_plt,
			      struct arm_plt_info *arm_plt,
			      int dynindx, bfd_vma sym_value)
{
  struct elf32_arm_link_hash_table *htab;
  asection *sgot;
  asection *splt;
  asection *srel;
  bfd_byte *loc;
  bfd_vma plt_index;
  Elf_Internal_Rela rel;
  bfd_vma plt_header_size;
  bfd_vma got_header_size;
 
  htab = elf32_arm_hash_table (info);
 
  /* Pick the appropriate sections and sizes.  */
  if (dynindx == -1)
    {
      splt = htab->root.iplt;
      sgot = htab->root.igotplt;
      srel = htab->root.irelplt;
 
      /* There are no reserved entries in .igot.plt, and no special
	 first entry in .iplt.  */
      got_header_size = 0;
      plt_header_size = 0;
    }
  else
    {
      splt = htab->root.splt;
      sgot = htab->root.sgotplt;
      srel = htab->root.srelplt;
 
      got_header_size = get_elf_backend_data (output_bfd)->got_header_size;
      plt_header_size = htab->plt_header_size;
    }
  BFD_ASSERT (splt != NULL && srel != NULL);
 
  /* Fill in the entry in the procedure linkage table.  */
  if (htab->symbian_p)
    {
      BFD_ASSERT (dynindx >= 0);
      put_arm_insn (htab, output_bfd,
		    elf32_arm_symbian_plt_entry[0],
		    splt->contents + root_plt->offset);
      bfd_put_32 (output_bfd,
		  elf32_arm_symbian_plt_entry[1],
		  splt->contents + root_plt->offset + 4);
 
      /* Fill in the entry in the .rel.plt section.  */
      rel.r_offset = (splt->output_section->vma
		      + splt->output_offset
		      + root_plt->offset + 4);
      rel.r_info = ELF32_R_INFO (dynindx, R_ARM_GLOB_DAT);
 
      /* Get the index in the procedure linkage table which
	 corresponds to this symbol.  This is the index of this symbol
	 in all the symbols for which we are making plt entries.  The
	 first entry in the procedure linkage table is reserved.  */
      plt_index = ((root_plt->offset - plt_header_size)
		   / htab->plt_entry_size);
    }
  else
    {
      bfd_vma got_offset, got_address, plt_address;
      bfd_vma got_displacement, initial_got_entry;
      bfd_byte * ptr;
 
      BFD_ASSERT (sgot != NULL);
 
      /* Get the offset into the .(i)got.plt table of the entry that
	 corresponds to this function.  */
      got_offset = (arm_plt->got_offset & -2);
 
      /* Get the index in the procedure linkage table which
	 corresponds to this symbol.  This is the index of this symbol
	 in all the symbols for which we are making plt entries.
	 After the reserved .got.plt entries, all symbols appear in
	 the same order as in .plt.  */
      plt_index = (got_offset - got_header_size) / 4;
 
      /* Calculate the address of the GOT entry.  */
      got_address = (sgot->output_section->vma
		     + sgot->output_offset
		     + got_offset);
 
      /* ...and the address of the PLT entry.  */
      plt_address = (splt->output_section->vma
		     + splt->output_offset
		     + root_plt->offset);
 
      ptr = splt->contents + root_plt->offset;
      if (htab->vxworks_p && info->shared)
	{
	  unsigned int i;
	  bfd_vma val;
 
	  for (i = 0; i != htab->plt_entry_size / 4; i++, ptr += 4)
	    {
	      val = elf32_arm_vxworks_shared_plt_entry[i];
	      if (i == 2)
		val |= got_address - sgot->output_section->vma;
	      if (i == 5)
		val |= plt_index * RELOC_SIZE (htab);
	      if (i == 2 || i == 5)
		bfd_put_32 (output_bfd, val, ptr);
	      else
		put_arm_insn (htab, output_bfd, val, ptr);
	    }
	}
      else if (htab->vxworks_p)
	{
	  unsigned int i;
	  bfd_vma val;
 
	  for (i = 0; i != htab->plt_entry_size / 4; i++, ptr += 4)
	    {
	      val = elf32_arm_vxworks_exec_plt_entry[i];
	      if (i == 2)
		val |= got_address;
	      if (i == 4)
		val |= 0xffffff & -((root_plt->offset + i * 4 + 8) >> 2);
	      if (i == 5)
		val |= plt_index * RELOC_SIZE (htab);
	      if (i == 2 || i == 5)
		bfd_put_32 (output_bfd, val, ptr);
	      else
		put_arm_insn (htab, output_bfd, val, ptr);
	    }
 
	  loc = (htab->srelplt2->contents
		 + (plt_index * 2 + 1) * RELOC_SIZE (htab));
 
	  /* Create the .rela.plt.unloaded R_ARM_ABS32 relocation
	     referencing the GOT for this PLT entry.  */
	  rel.r_offset = plt_address + 8;
	  rel.r_info = ELF32_R_INFO (htab->root.hgot->indx, R_ARM_ABS32);
	  rel.r_addend = got_offset;
	  SWAP_RELOC_OUT (htab) (output_bfd, &rel, loc);
	  loc += RELOC_SIZE (htab);
 
	  /* Create the R_ARM_ABS32 relocation referencing the
	     beginning of the PLT for this GOT entry.  */
	  rel.r_offset = got_address;
	  rel.r_info = ELF32_R_INFO (htab->root.hplt->indx, R_ARM_ABS32);
	  rel.r_addend = 0;
	  SWAP_RELOC_OUT (htab) (output_bfd, &rel, loc);
	}
      else
	{
	  /* Calculate the displacement between the PLT slot and the
	     entry in the GOT.  The eight-byte offset accounts for the
	     value produced by adding to pc in the first instruction
	     of the PLT stub.  */
	  got_displacement = got_address - (plt_address + 8);
 
	  BFD_ASSERT ((got_displacement & 0xf0000000) == 0);
 
	  if (elf32_arm_plt_needs_thumb_stub_p (info, arm_plt))
	    {
	      put_thumb_insn (htab, output_bfd,
			      elf32_arm_plt_thumb_stub[0], ptr - 4);
	      put_thumb_insn (htab, output_bfd,
			      elf32_arm_plt_thumb_stub[1], ptr - 2);
	    }
 
	  put_arm_insn (htab, output_bfd,
			elf32_arm_plt_entry[0]
			| ((got_displacement & 0x0ff00000) >> 20),
			ptr + 0);
	  put_arm_insn (htab, output_bfd,
			elf32_arm_plt_entry[1]
			| ((got_displacement & 0x000ff000) >> 12),
			ptr+ 4);
	  put_arm_insn (htab, output_bfd,
			elf32_arm_plt_entry[2]
			| (got_displacement & 0x00000fff),
			ptr + 8);
#ifdef FOUR_WORD_PLT
	  bfd_put_32 (output_bfd, elf32_arm_plt_entry[3], ptr + 12);
#endif
	}
 
      /* Fill in the entry in the .rel(a).(i)plt section.  */
      rel.r_offset = got_address;
      rel.r_addend = 0;
      if (dynindx == -1)
	{
	  /* .igot.plt entries use IRELATIVE relocations against SYM_VALUE.
	     The dynamic linker or static executable then calls SYM_VALUE
	     to determine the correct run-time value of the .igot.plt entry.  */
	  rel.r_info = ELF32_R_INFO (0, R_ARM_IRELATIVE);
	  initial_got_entry = sym_value;
	}
      else
	{
	  rel.r_info = ELF32_R_INFO (dynindx, R_ARM_JUMP_SLOT);
	  initial_got_entry = (splt->output_section->vma
			       + splt->output_offset);
	}
 
      /* Fill in the entry in the global offset table.  */
      bfd_put_32 (output_bfd, initial_got_entry,
		  sgot->contents + got_offset);
    }
 
  loc = srel->contents + plt_index * RELOC_SIZE (htab);
  SWAP_RELOC_OUT (htab) (output_bfd, &rel, loc);
}
 
/* Some relocations map to different relocations depending on the
   target.  Return the real relocation.  */
 
static int
arm_real_reloc_type (struct elf32_arm_link_hash_table * globals,
		     int r_type)
{
  switch (r_type)
    {
    case R_ARM_TARGET1:
      if (globals->target1_is_rel)
	return R_ARM_REL32;
      else
	return R_ARM_ABS32;
 
    case R_ARM_TARGET2:
      return globals->target2_reloc;
 
    default:
      return r_type;
    }
}
 
/* Return the base VMA address which should be subtracted from real addresses
   when resolving @dtpoff relocation.
   This is PT_TLS segment p_vaddr.  */
 
static bfd_vma
dtpoff_base (struct bfd_link_info *info)
{
  /* If tls_sec is NULL, we should have signalled an error already.  */
  if (elf_hash_table (info)->tls_sec == NULL)
    return 0;
  return elf_hash_table (info)->tls_sec->vma;
}
 
/* Return the relocation value for @tpoff relocation
   if STT_TLS virtual address is ADDRESS.  */
 
static bfd_vma
tpoff (struct bfd_link_info *info, bfd_vma address)
{
  struct elf_link_hash_table *htab = elf_hash_table (info);
  bfd_vma base;
 
  /* If tls_sec is NULL, we should have signalled an error already.  */
  if (htab->tls_sec == NULL)
    return 0;
  base = align_power ((bfd_vma) TCB_SIZE, htab->tls_sec->alignment_power);
  return address - htab->tls_sec->vma + base;
}
 
/* Perform an R_ARM_ABS12 relocation on the field pointed to by DATA.
   VALUE is the relocation value.  */
 
static bfd_reloc_status_type
elf32_arm_abs12_reloc (bfd *abfd, void *data, bfd_vma value)
{
  if (value > 0xfff)
    return bfd_reloc_overflow;
 
  value |= bfd_get_32 (abfd, data) & 0xfffff000;
  bfd_put_32 (abfd, value, data);
  return bfd_reloc_ok;
}
 
/* Handle TLS relaxations.  Relaxing is possible for symbols that use
   R_ARM_GOTDESC, R_ARM_{,THM_}TLS_CALL or
   R_ARM_{,THM_}TLS_DESCSEQ relocations, during a static link.
 
   Return bfd_reloc_ok if we're done, bfd_reloc_continue if the caller
   is to then call final_link_relocate.  Return other values in the
   case of error.
 
   FIXME:When --emit-relocs is in effect, we'll emit relocs describing
   the pre-relaxed code.  It would be nice if the relocs were updated
   to match the optimization.   */
 
static bfd_reloc_status_type 
elf32_arm_tls_relax (struct elf32_arm_link_hash_table *globals,
		     bfd *input_bfd, asection *input_sec, bfd_byte *contents, 
		     Elf_Internal_Rela *rel, unsigned long is_local)
{
  unsigned long insn;
 
  switch (ELF32_R_TYPE (rel->r_info))
    {
    default:
      return bfd_reloc_notsupported;
 
    case R_ARM_TLS_GOTDESC:
      if (is_local)
	insn = 0;
      else
	{
	  insn = bfd_get_32 (input_bfd, contents + rel->r_offset);
	  if (insn & 1)
	    insn -= 5; /* THUMB */
	  else
	    insn -= 8; /* ARM */
	}
      bfd_put_32 (input_bfd, insn, contents + rel->r_offset);
      return bfd_reloc_continue;
 
    case R_ARM_THM_TLS_DESCSEQ:
      /* Thumb insn.  */
      insn = bfd_get_16 (input_bfd, contents + rel->r_offset);
      if ((insn & 0xff78) == 0x4478)	  /* add rx, pc */
	{
	  if (is_local)
	    /* nop */
	    bfd_put_16 (input_bfd, 0x46c0, contents + rel->r_offset);
	}
      else if ((insn & 0xffc0) == 0x6840)  /* ldr rx,[ry,#4] */
	{
	  if (is_local)
	    /* nop */
	    bfd_put_16 (input_bfd, 0x46c0, contents + rel->r_offset);
	  else
	    /* ldr rx,[ry] */
	    bfd_put_16 (input_bfd, insn & 0xf83f, contents + rel->r_offset);
	}
      else if ((insn & 0xff87) == 0x4780)  /* blx rx */
	{
	  if (is_local)
	    /* nop */
	    bfd_put_16 (input_bfd, 0x46c0, contents + rel->r_offset);
	  else
	    /* mov r0, rx */
	    bfd_put_16 (input_bfd, 0x4600 | (insn & 0x78),
			contents + rel->r_offset);
	}
      else
	{
	  if ((insn & 0xf000) == 0xf000 || (insn & 0xf800) == 0xe800)
	    /* It's a 32 bit instruction, fetch the rest of it for
	       error generation.  */
	    insn = (insn << 16)
	      | bfd_get_16 (input_bfd, contents + rel->r_offset + 2);
	  (*_bfd_error_handler)
	    (_("%B(%A+0x%lx):unexpected Thumb instruction '0x%x' in TLS trampoline"),
	     input_bfd, input_sec, (unsigned long)rel->r_offset, insn);
	  return bfd_reloc_notsupported;
	}
      break;
 
    case R_ARM_TLS_DESCSEQ:
      /* arm insn.  */
      insn = bfd_get_32 (input_bfd, contents + rel->r_offset);
      if ((insn & 0xffff0ff0) == 0xe08f0000) /* add rx,pc,ry */
	{
	  if (is_local)
	    /* mov rx, ry */
	    bfd_put_32 (input_bfd, 0xe1a00000 | (insn & 0xffff),
			contents + rel->r_offset);
	}
      else if ((insn & 0xfff00fff) == 0xe5900004) /* ldr rx,[ry,#4]*/
	{
	  if (is_local)
	    /* nop */
	    bfd_put_32 (input_bfd, 0xe1a00000, contents + rel->r_offset);
	  else
	    /* ldr rx,[ry] */
	    bfd_put_32 (input_bfd, insn & 0xfffff000,
			contents + rel->r_offset);
	}
      else if ((insn & 0xfffffff0) == 0xe12fff30) /* blx rx */
	{
	  if (is_local)
	    /* nop */
	    bfd_put_32 (input_bfd, 0xe1a00000, contents + rel->r_offset);
	  else
	    /* mov r0, rx */
	    bfd_put_32 (input_bfd, 0xe1a00000 | (insn & 0xf),
			contents + rel->r_offset);
	}
      else
	{
	  (*_bfd_error_handler)
	    (_("%B(%A+0x%lx):unexpected ARM instruction '0x%x' in TLS trampoline"),
	     input_bfd, input_sec, (unsigned long)rel->r_offset, insn);
	  return bfd_reloc_notsupported;
	}
      break;
 
    case R_ARM_TLS_CALL:
      /* GD->IE relaxation, turn the instruction into 'nop' or
	 'ldr r0, [pc,r0]'  */
      insn = is_local ? 0xe1a00000 : 0xe79f0000;
      bfd_put_32 (input_bfd, insn, contents + rel->r_offset);
      break;
 
    case R_ARM_THM_TLS_CALL:
      /* GD->IE relaxation */
      if (!is_local)
	/* add r0,pc; ldr r0, [r0]  */
	insn = 0x44786800;
      else if (arch_has_thumb2_nop (globals))
	/* nop.w */
	insn = 0xf3af8000;
      else
	/* nop; nop */
	insn = 0xbf00bf00;
 
      bfd_put_16 (input_bfd, insn >> 16, contents + rel->r_offset);
      bfd_put_16 (input_bfd, insn & 0xffff, contents + rel->r_offset + 2);
      break;
    }
  return bfd_reloc_ok;
}
 
/* For a given value of n, calculate the value of G_n as required to
   deal with group relocations.  We return it in the form of an
   encoded constant-and-rotation, together with the final residual.  If n is
   specified as less than zero, then final_residual is filled with the
   input value and no further action is performed.  */
 
static bfd_vma
calculate_group_reloc_mask (bfd_vma value, int n, bfd_vma *final_residual)
{
  int current_n;
  bfd_vma g_n;
  bfd_vma encoded_g_n = 0;
  bfd_vma residual = value; /* Also known as Y_n.  */
 
  for (current_n = 0; current_n <= n; current_n++)
    {
      int shift;
 
      /* Calculate which part of the value to mask.  */
      if (residual == 0)
        shift = 0;
      else
        {
          int msb;
 
          /* Determine the most significant bit in the residual and
             align the resulting value to a 2-bit boundary.  */
          for (msb = 30; msb >= 0; msb -= 2)
            if (residual & (3 << msb))
              break;
 
          /* The desired shift is now (msb - 6), or zero, whichever
             is the greater.  */
          shift = msb - 6;
          if (shift < 0)
            shift = 0;
        }
 
      /* Calculate g_n in 32-bit as well as encoded constant+rotation form.  */
      g_n = residual & (0xff << shift);
      encoded_g_n = (g_n >> shift)
                    | ((g_n <= 0xff ? 0 : (32 - shift) / 2) << 8);
 
      /* Calculate the residual for the next time around.  */
      residual &= ~g_n;
    }
 
  *final_residual = residual;
 
  return encoded_g_n;
}
 
/* Given an ARM instruction, determine whether it is an ADD or a SUB.
   Returns 1 if it is an ADD, -1 if it is a SUB, and 0 otherwise.  */
 
static int
identify_add_or_sub (bfd_vma insn)
{
  int opcode = insn & 0x1e00000;
 
  if (opcode == 1 << 23) /* ADD */
    return 1;
 
  if (opcode == 1 << 22) /* SUB */
    return -1;
 
  return 0;
}
 
/* Perform a relocation as part of a final link.  */
 
static bfd_reloc_status_type
elf32_arm_final_link_relocate (reloc_howto_type *           howto,
			       bfd *                        input_bfd,
			       bfd *                        output_bfd,
			       asection *                   input_section,
			       bfd_byte *                   contents,
			       Elf_Internal_Rela *          rel,
			       bfd_vma                      value,
			       struct bfd_link_info *       info,
			       asection *                   sym_sec,
			       const char *                 sym_name,
			       unsigned char                st_type,
			       enum arm_st_branch_type      branch_type,
			       struct elf_link_hash_entry * h,
			       bfd_boolean *                unresolved_reloc_p,
			       char **                      error_message)
{
  unsigned long                 r_type = howto->type;
  unsigned long                 r_symndx;
  bfd_byte *                    hit_data = contents + rel->r_offset;
  bfd_vma *                     local_got_offsets;
  bfd_vma *                     local_tlsdesc_gotents;
  asection *                    sgot;
  asection *                    splt;
  asection *                    sreloc = NULL;
  asection *                    srelgot;
  bfd_vma                       addend;
  bfd_signed_vma                signed_addend;
  unsigned char                 dynreloc_st_type;
  bfd_vma                       dynreloc_value;
  struct elf32_arm_link_hash_table * globals;
  struct elf32_arm_link_hash_entry *eh;
  union gotplt_union           *root_plt;
  struct arm_plt_info          *arm_plt;
  bfd_vma                       plt_offset;
  bfd_vma                       gotplt_offset;
  bfd_boolean                   has_iplt_entry;
 
  globals = elf32_arm_hash_table (info);
  if (globals == NULL)
    return bfd_reloc_notsupported;
 
  BFD_ASSERT (is_arm_elf (input_bfd));
 
  /* Some relocation types map to different relocations depending on the
     target.  We pick the right one here.  */
  r_type = arm_real_reloc_type (globals, r_type);
 
  /* It is possible to have linker relaxations on some TLS access
     models.  Update our information here.  */
  r_type = elf32_arm_tls_transition (info, r_type, h);
 
  if (r_type != howto->type)
    howto = elf32_arm_howto_from_type (r_type);
 
  /* If the start address has been set, then set the EF_ARM_HASENTRY
     flag.  Setting this more than once is redundant, but the cost is
     not too high, and it keeps the code simple.
 
     The test is done  here, rather than somewhere else, because the
     start address is only set just before the final link commences.
 
     Note - if the user deliberately sets a start address of 0, the
     flag will not be set.  */
  if (bfd_get_start_address (output_bfd) != 0)
    elf_elfheader (output_bfd)->e_flags |= EF_ARM_HASENTRY;
 
  eh = (struct elf32_arm_link_hash_entry *) h;
  sgot = globals->root.sgot;
  local_got_offsets = elf_local_got_offsets (input_bfd);
  local_tlsdesc_gotents = elf32_arm_local_tlsdesc_gotent (input_bfd);
 
  if (globals->root.dynamic_sections_created)
    srelgot = globals->root.srelgot;
  else
    srelgot = NULL;
 
  r_symndx = ELF32_R_SYM (rel->r_info);
 
  if (globals->use_rel)
    {
      addend = bfd_get_32 (input_bfd, hit_data) & howto->src_mask;
 
      if (addend & ((howto->src_mask + 1) >> 1))
	{
	  signed_addend = -1;
	  signed_addend &= ~ howto->src_mask;
	  signed_addend |= addend;
	}
      else
	signed_addend = addend;
    }
  else
    addend = signed_addend = rel->r_addend;
 
  /* Record the symbol information that should be used in dynamic
     relocations.  */
  dynreloc_st_type = st_type;
  dynreloc_value = value;
  if (branch_type == ST_BRANCH_TO_THUMB)
    dynreloc_value |= 1;
 
  /* Find out whether the symbol has a PLT.  Set ST_VALUE, BRANCH_TYPE and
     VALUE appropriately for relocations that we resolve at link time.  */
  has_iplt_entry = FALSE;
  if (elf32_arm_get_plt_info (input_bfd, eh, r_symndx, &root_plt, &arm_plt)
      && root_plt->offset != (bfd_vma) -1)
    {
      plt_offset = root_plt->offset;
      gotplt_offset = arm_plt->got_offset;
 
      if (h == NULL || eh->is_iplt)
	{
	  has_iplt_entry = TRUE;
	  splt = globals->root.iplt;
 
	  /* Populate .iplt entries here, because not all of them will
	     be seen by finish_dynamic_symbol.  The lower bit is set if
	     we have already populated the entry.  */
	  if (plt_offset & 1)
	    plt_offset--;
	  else
	    {
	      elf32_arm_populate_plt_entry (output_bfd, info, root_plt, arm_plt,
					    -1, dynreloc_value);
	      root_plt->offset |= 1;
	    }
 
	  /* Static relocations always resolve to the .iplt entry.  */
	  st_type = STT_FUNC;
	  value = (splt->output_section->vma
		   + splt->output_offset
		   + plt_offset);
	  branch_type = ST_BRANCH_TO_ARM;
 
	  /* If there are non-call relocations that resolve to the .iplt
	     entry, then all dynamic ones must too.  */
	  if (arm_plt->noncall_refcount != 0)
	    {
	      dynreloc_st_type = st_type;
	      dynreloc_value = value;
	    }
	}
      else
	/* We populate the .plt entry in finish_dynamic_symbol.  */
	splt = globals->root.splt;
    }
  else
    {
      splt = NULL;
      plt_offset = (bfd_vma) -1;
      gotplt_offset = (bfd_vma) -1;
    }
 
  switch (r_type)
    {
    case R_ARM_NONE:
      /* We don't need to find a value for this symbol.  It's just a
	 marker.  */
      *unresolved_reloc_p = FALSE;
      return bfd_reloc_ok;
 
    case R_ARM_ABS12:
      if (!globals->vxworks_p)
	return elf32_arm_abs12_reloc (input_bfd, hit_data, value + addend);
 
    case R_ARM_PC24:
    case R_ARM_ABS32:
    case R_ARM_ABS32_NOI:
    case R_ARM_REL32:
    case R_ARM_REL32_NOI:
    case R_ARM_CALL:
    case R_ARM_JUMP24:
    case R_ARM_XPC25:
    case R_ARM_PREL31:
    case R_ARM_PLT32:
      /* Handle relocations which should use the PLT entry.  ABS32/REL32
	 will use the symbol's value, which may point to a PLT entry, but we
	 don't need to handle that here.  If we created a PLT entry, all
	 branches in this object should go to it, except if the PLT is too
	 far away, in which case a long branch stub should be inserted.  */
      if ((r_type != R_ARM_ABS32 && r_type != R_ARM_REL32
           && r_type != R_ARM_ABS32_NOI && r_type != R_ARM_REL32_NOI
	   && r_type != R_ARM_CALL
	   && r_type != R_ARM_JUMP24
	   && r_type != R_ARM_PLT32)
	  && plt_offset != (bfd_vma) -1)
	{
	  /* If we've created a .plt section, and assigned a PLT entry
	     to this function, it must either be a STT_GNU_IFUNC reference
	     or not be known to bind locally.  In other cases, we should
	     have cleared the PLT entry by now.  */
	  BFD_ASSERT (has_iplt_entry || !SYMBOL_CALLS_LOCAL (info, h));
 
	  value = (splt->output_section->vma
		   + splt->output_offset
		   + plt_offset);
	  *unresolved_reloc_p = FALSE;
	  return _bfd_final_link_relocate (howto, input_bfd, input_section,
					   contents, rel->r_offset, value,
					   rel->r_addend);
	}
 
      /* When generating a shared object or relocatable executable, these
	 relocations are copied into the output file to be resolved at
	 run time.  */
      if ((info->shared || globals->root.is_relocatable_executable)
	  && (input_section->flags & SEC_ALLOC)
	  && !(globals->vxworks_p
	       && strcmp (input_section->output_section->name,
			  ".tls_vars") == 0)
	  && ((r_type != R_ARM_REL32 && r_type != R_ARM_REL32_NOI)
	      || !SYMBOL_CALLS_LOCAL (info, h))
	  && (!strstr (input_section->name, STUB_SUFFIX))
	  && (h == NULL
	      || ELF_ST_VISIBILITY (h->other) == STV_DEFAULT
	      || h->root.type != bfd_link_hash_undefweak)
	  && r_type != R_ARM_PC24
	  && r_type != R_ARM_CALL
	  && r_type != R_ARM_JUMP24
	  && r_type != R_ARM_PREL31
	  && r_type != R_ARM_PLT32)
	{
	  Elf_Internal_Rela outrel;
	  bfd_boolean skip, relocate;
 
	  *unresolved_reloc_p = FALSE;
 
	  if (sreloc == NULL && globals->root.dynamic_sections_created)
	    {
	      sreloc = _bfd_elf_get_dynamic_reloc_section (input_bfd, input_section,
							   ! globals->use_rel);
 
	      if (sreloc == NULL)
		return bfd_reloc_notsupported;
	    }
 
	  skip = FALSE;
	  relocate = FALSE;
 
	  outrel.r_addend = addend;
	  outrel.r_offset =
	    _bfd_elf_section_offset (output_bfd, info, input_section,
				     rel->r_offset);
	  if (outrel.r_offset == (bfd_vma) -1)
	    skip = TRUE;
	  else if (outrel.r_offset == (bfd_vma) -2)
	    skip = TRUE, relocate = TRUE;
	  outrel.r_offset += (input_section->output_section->vma
			      + input_section->output_offset);
 
	  if (skip)
	    memset (&outrel, 0, sizeof outrel);
	  else if (h != NULL
		   && h->dynindx != -1
		   && (!info->shared
		       || !info->symbolic
		       || !h->def_regular))
	    outrel.r_info = ELF32_R_INFO (h->dynindx, r_type);
	  else
	    {
	      int symbol;
 
	      /* This symbol is local, or marked to become local.  */
	      BFD_ASSERT (r_type == R_ARM_ABS32 || r_type == R_ARM_ABS32_NOI);
	      if (globals->symbian_p)
		{
		  asection *osec;
 
		  /* On Symbian OS, the data segment and text segement
		     can be relocated independently.  Therefore, we
		     must indicate the segment to which this
		     relocation is relative.  The BPABI allows us to
		     use any symbol in the right segment; we just use
		     the section symbol as it is convenient.  (We
		     cannot use the symbol given by "h" directly as it
		     will not appear in the dynamic symbol table.)
 
		     Note that the dynamic linker ignores the section
		     symbol value, so we don't subtract osec->vma
		     from the emitted reloc addend.  */
		  if (sym_sec)
		    osec = sym_sec->output_section;
		  else
		    osec = input_section->output_section;
		  symbol = elf_section_data (osec)->dynindx;
		  if (symbol == 0)
		    {
		      struct elf_link_hash_table *htab = elf_hash_table (info);
 
		      if ((osec->flags & SEC_READONLY) == 0
			  && htab->data_index_section != NULL)
			osec = htab->data_index_section;
		      else
			osec = htab->text_index_section;
		      symbol = elf_section_data (osec)->dynindx;
		    }
		  BFD_ASSERT (symbol != 0);
		}
	      else
		/* On SVR4-ish systems, the dynamic loader cannot
		   relocate the text and data segments independently,
		   so the symbol does not matter.  */
		symbol = 0;
	      if (dynreloc_st_type == STT_GNU_IFUNC)
		/* We have an STT_GNU_IFUNC symbol that doesn't resolve
		   to the .iplt entry.  Instead, every non-call reference
		   must use an R_ARM_IRELATIVE relocation to obtain the
		   correct run-time address.  */
		outrel.r_info = ELF32_R_INFO (symbol, R_ARM_IRELATIVE);
	      else
		outrel.r_info = ELF32_R_INFO (symbol, R_ARM_RELATIVE);
	      if (globals->use_rel)
		relocate = TRUE;
	      else
		outrel.r_addend += dynreloc_value;
	    }
 
	  elf32_arm_add_dynreloc (output_bfd, info, sreloc, &outrel);
 
	  /* If this reloc is against an external symbol, we do not want to
	     fiddle with the addend.  Otherwise, we need to include the symbol
	     value so that it becomes an addend for the dynamic reloc.  */
	  if (! relocate)
	    return bfd_reloc_ok;
 
	  return _bfd_final_link_relocate (howto, input_bfd, input_section,
					   contents, rel->r_offset,
					   dynreloc_value, (bfd_vma) 0);
	}
      else switch (r_type)
	{
	case R_ARM_ABS12:
	  return elf32_arm_abs12_reloc (input_bfd, hit_data, value + addend);
 
	case R_ARM_XPC25:	  /* Arm BLX instruction.  */
	case R_ARM_CALL:
	case R_ARM_JUMP24:
	case R_ARM_PC24:	  /* Arm B/BL instruction.  */
	case R_ARM_PLT32:
	  {
	  struct elf32_arm_stub_hash_entry *stub_entry = NULL;
 
	  if (r_type == R_ARM_XPC25)
	    {
	      /* Check for Arm calling Arm function.  */
	      /* FIXME: Should we translate the instruction into a BL
		 instruction instead ?  */
	      if (branch_type != ST_BRANCH_TO_THUMB)
		(*_bfd_error_handler)
		  (_("\%B: Warning: Arm BLX instruction targets Arm function '%s'."),
		   input_bfd,
		   h ? h->root.root.string : "(local)");
	    }
	  else if (r_type == R_ARM_PC24)
	    {
	      /* Check for Arm calling Thumb function.  */
	      if (branch_type == ST_BRANCH_TO_THUMB)
		{
		  if (elf32_arm_to_thumb_stub (info, sym_name, input_bfd,
					       output_bfd, input_section,
					       hit_data, sym_sec, rel->r_offset,
					       signed_addend, value,
					       error_message))
		    return bfd_reloc_ok;
		  else
		    return bfd_reloc_dangerous;
		}
	    }
 
	  /* Check if a stub has to be inserted because the
	     destination is too far or we are changing mode.  */
	  if (   r_type == R_ARM_CALL
	      || r_type == R_ARM_JUMP24
	      || r_type == R_ARM_PLT32)
	    {
	      enum elf32_arm_stub_type stub_type = arm_stub_none;
	      struct elf32_arm_link_hash_entry *hash;
 
	      hash = (struct elf32_arm_link_hash_entry *) h;
	      stub_type = arm_type_of_stub (info, input_section, rel,
					    st_type, &branch_type,
					    hash, value, sym_sec,
					    input_bfd, sym_name);
 
	      if (stub_type != arm_stub_none)
		{
		  /* The target is out of reach, so redirect the
		     branch to the local stub for this function.  */
 
		  stub_entry = elf32_arm_get_stub_entry (input_section,
							 sym_sec, h,
							 rel, globals,
							 stub_type);
		  if (stub_entry != NULL)
		    value = (stub_entry->stub_offset
			     + stub_entry->stub_sec->output_offset
			     + stub_entry->stub_sec->output_section->vma);
		}
	      else
		{
		  /* If the call goes through a PLT entry, make sure to
		     check distance to the right destination address.  */
		  if (plt_offset != (bfd_vma) -1)
		    {
		      value = (splt->output_section->vma
			       + splt->output_offset
			       + plt_offset);
		      *unresolved_reloc_p = FALSE;
		      /* The PLT entry is in ARM mode, regardless of the
			 target function.  */
		      branch_type = ST_BRANCH_TO_ARM;
		    }
		}
	    }
 
	  /* The ARM ELF ABI says that this reloc is computed as: S - P + A
	     where:
	      S is the address of the symbol in the relocation.
	      P is address of the instruction being relocated.
	      A is the addend (extracted from the instruction) in bytes.
 
	     S is held in 'value'.
	     P is the base address of the section containing the
	       instruction plus the offset of the reloc into that
	       section, ie:
		 (input_section->output_section->vma +
		  input_section->output_offset +
		  rel->r_offset).
	     A is the addend, converted into bytes, ie:
		 (signed_addend * 4)
 
	     Note: None of these operations have knowledge of the pipeline
	     size of the processor, thus it is up to the assembler to
	     encode this information into the addend.  */
	  value -= (input_section->output_section->vma
		    + input_section->output_offset);
	  value -= rel->r_offset;
	  if (globals->use_rel)
	    value += (signed_addend << howto->size);
	  else
	    /* RELA addends do not have to be adjusted by howto->size.  */
	    value += signed_addend;
 
	  signed_addend = value;
	  signed_addend >>= howto->rightshift;
 
	  /* A branch to an undefined weak symbol is turned into a jump to
	     the next instruction unless a PLT entry will be created.
	     Do the same for local undefined symbols (but not for STN_UNDEF).
	     The jump to the next instruction is optimized as a NOP depending
	     on the architecture.  */
	  if (h ? (h->root.type == bfd_link_hash_undefweak
		   && plt_offset == (bfd_vma) -1)
	      : r_symndx != STN_UNDEF && bfd_is_und_section (sym_sec))
	    {
	      value = (bfd_get_32 (input_bfd, hit_data) & 0xf0000000);
 
	      if (arch_has_arm_nop (globals))
		value |= 0x0320f000;
	      else
		value |= 0x01a00000; /* Using pre-UAL nop: mov r0, r0.  */
	    }
	  else
	    {
	      /* Perform a signed range check.  */
	      if (   signed_addend >   ((bfd_signed_vma)  (howto->dst_mask >> 1))
		  || signed_addend < - ((bfd_signed_vma) ((howto->dst_mask + 1) >> 1)))
		return bfd_reloc_overflow;
 
	      addend = (value & 2);
 
	      value = (signed_addend & howto->dst_mask)
		| (bfd_get_32 (input_bfd, hit_data) & (~ howto->dst_mask));
 
	      if (r_type == R_ARM_CALL)
		{
		  /* Set the H bit in the BLX instruction.  */
		  if (branch_type == ST_BRANCH_TO_THUMB)
		    {
		      if (addend)
			value |= (1 << 24);
		      else
			value &= ~(bfd_vma)(1 << 24);
		    }
 
		  /* Select the correct instruction (BL or BLX).  */
		  /* Only if we are not handling a BL to a stub. In this
		     case, mode switching is performed by the stub.  */
		  if (branch_type == ST_BRANCH_TO_THUMB && !stub_entry)
		    value |= (1 << 28);
		  else if (stub_entry || branch_type != ST_BRANCH_UNKNOWN)
		    {
		      value &= ~(bfd_vma)(1 << 28);
		      value |= (1 << 24);
		    }
		}
	    }
	  }
	  break;
 
	case R_ARM_ABS32:
	  value += addend;
	  if (branch_type == ST_BRANCH_TO_THUMB)
	    value |= 1;
	  break;
 
	case R_ARM_ABS32_NOI:
	  value += addend;
	  break;
 
	case R_ARM_REL32:
	  value += addend;
	  if (branch_type == ST_BRANCH_TO_THUMB)
	    value |= 1;
	  value -= (input_section->output_section->vma
		    + input_section->output_offset + rel->r_offset);
	  break;
 
	case R_ARM_REL32_NOI:
	  value += addend;
	  value -= (input_section->output_section->vma
		    + input_section->output_offset + rel->r_offset);
	  break;
 
	case R_ARM_PREL31:
	  value -= (input_section->output_section->vma
		    + input_section->output_offset + rel->r_offset);
	  value += signed_addend;
	  if (! h || h->root.type != bfd_link_hash_undefweak)
	    {
	      /* Check for overflow.  */
	      if ((value ^ (value >> 1)) & (1 << 30))
		return bfd_reloc_overflow;
	    }
	  value &= 0x7fffffff;
	  value |= (bfd_get_32 (input_bfd, hit_data) & 0x80000000);
	  if (branch_type == ST_BRANCH_TO_THUMB)
	    value |= 1;
	  break;
	}
 
      bfd_put_32 (input_bfd, value, hit_data);
      return bfd_reloc_ok;
 
    case R_ARM_ABS8:
      value += addend;
 
      /* There is no way to tell whether the user intended to use a signed or
	 unsigned addend.  When checking for overflow we accept either,
	 as specified by the AAELF.  */
      if ((long) value > 0xff || (long) value < -0x80)
	return bfd_reloc_overflow;
 
      bfd_put_8 (input_bfd, value, hit_data);
      return bfd_reloc_ok;
 
    case R_ARM_ABS16:
      value += addend;
 
      /* See comment for R_ARM_ABS8.  */
      if ((long) value > 0xffff || (long) value < -0x8000)
	return bfd_reloc_overflow;
 
      bfd_put_16 (input_bfd, value, hit_data);
      return bfd_reloc_ok;
 
    case R_ARM_THM_ABS5:
      /* Support ldr and str instructions for the thumb.  */
      if (globals->use_rel)
	{
	  /* Need to refetch addend.  */
	  addend = bfd_get_16 (input_bfd, hit_data) & howto->src_mask;
	  /* ??? Need to determine shift amount from operand size.  */
	  addend >>= howto->rightshift;
	}
      value += addend;
 
      /* ??? Isn't value unsigned?  */
      if ((long) value > 0x1f || (long) value < -0x10)
	return bfd_reloc_overflow;
 
      /* ??? Value needs to be properly shifted into place first.  */
      value |= bfd_get_16 (input_bfd, hit_data) & 0xf83f;
      bfd_put_16 (input_bfd, value, hit_data);
      return bfd_reloc_ok;
 
    case R_ARM_THM_ALU_PREL_11_0:
      /* Corresponds to: addw.w reg, pc, #offset (and similarly for subw).  */
      {
	bfd_vma insn;
	bfd_signed_vma relocation;
 
	insn = (bfd_get_16 (input_bfd, hit_data) << 16)
             | bfd_get_16 (input_bfd, hit_data + 2);
 
        if (globals->use_rel)
          {
            signed_addend = (insn & 0xff) | ((insn & 0x7000) >> 4)
                          | ((insn & (1 << 26)) >> 15);
            if (insn & 0xf00000)
              signed_addend = -signed_addend;
          }
 
	relocation = value + signed_addend;
	relocation -= (input_section->output_section->vma
		       + input_section->output_offset
		       + rel->r_offset);
 
        value = abs (relocation);
 
        if (value >= 0x1000)
          return bfd_reloc_overflow;
 
	insn = (insn & 0xfb0f8f00) | (value & 0xff)
             | ((value & 0x700) << 4)
             | ((value & 0x800) << 15);
        if (relocation < 0)
          insn |= 0xa00000;
 
	bfd_put_16 (input_bfd, insn >> 16, hit_data);
	bfd_put_16 (input_bfd, insn & 0xffff, hit_data + 2);
 
        return bfd_reloc_ok;
      }
 
    case R_ARM_THM_PC8:
      /* PR 10073:  This reloc is not generated by the GNU toolchain,
	 but it is supported for compatibility with third party libraries
	 generated by other compilers, specifically the ARM/IAR.  */
      {
	bfd_vma insn;
	bfd_signed_vma relocation;
 
	insn = bfd_get_16 (input_bfd, hit_data);
 
        if (globals->use_rel)
	  addend = (insn & 0x00ff) << 2;
 
	relocation = value + addend;
	relocation -= (input_section->output_section->vma
		       + input_section->output_offset
		       + rel->r_offset);
 
        value = abs (relocation);
 
	/* We do not check for overflow of this reloc.  Although strictly
	   speaking this is incorrect, it appears to be necessary in order
	   to work with IAR generated relocs.  Since GCC and GAS do not
	   generate R_ARM_THM_PC8 relocs, the lack of a check should not be
	   a problem for them.  */
	value &= 0x3fc;
 
	insn = (insn & 0xff00) | (value >> 2);
 
	bfd_put_16 (input_bfd, insn, hit_data);
 
        return bfd_reloc_ok;
      }
 
    case R_ARM_THM_PC12:
      /* Corresponds to: ldr.w reg, [pc, #offset].  */
      {
	bfd_vma insn;
	bfd_signed_vma relocation;
 
	insn = (bfd_get_16 (input_bfd, hit_data) << 16)
             | bfd_get_16 (input_bfd, hit_data + 2);
 
        if (globals->use_rel)
          {
            signed_addend = insn & 0xfff;
            if (!(insn & (1 << 23)))
              signed_addend = -signed_addend;
          }
 
	relocation = value + signed_addend;
	relocation -= (input_section->output_section->vma
		       + input_section->output_offset
		       + rel->r_offset);
 
        value = abs (relocation);
 
        if (value >= 0x1000)
          return bfd_reloc_overflow;
 
	insn = (insn & 0xff7ff000) | value;
        if (relocation >= 0)
          insn |= (1 << 23);
 
	bfd_put_16 (input_bfd, insn >> 16, hit_data);
	bfd_put_16 (input_bfd, insn & 0xffff, hit_data + 2);
 
        return bfd_reloc_ok;
      }
 
    case R_ARM_THM_XPC22:
    case R_ARM_THM_CALL:
    case R_ARM_THM_JUMP24:
      /* Thumb BL (branch long instruction).  */
      {
	bfd_vma relocation;
        bfd_vma reloc_sign;
	bfd_boolean overflow = FALSE;
	bfd_vma upper_insn = bfd_get_16 (input_bfd, hit_data);
	bfd_vma lower_insn = bfd_get_16 (input_bfd, hit_data + 2);
	bfd_signed_vma reloc_signed_max;
	bfd_signed_vma reloc_signed_min;
	bfd_vma check;
	bfd_signed_vma signed_check;
	int bitsize;
	const int thumb2 = using_thumb2 (globals);
 
	/* A branch to an undefined weak symbol is turned into a jump to
	   the next instruction unless a PLT entry will be created.
	   The jump to the next instruction is optimized as a NOP.W for
	   Thumb-2 enabled architectures.  */
	if (h && h->root.type == bfd_link_hash_undefweak
	    && plt_offset == (bfd_vma) -1)
	  {
	    if (arch_has_thumb2_nop (globals))
	      {
		bfd_put_16 (input_bfd, 0xf3af, hit_data);
		bfd_put_16 (input_bfd, 0x8000, hit_data + 2);
	      }
	    else
	      {
		bfd_put_16 (input_bfd, 0xe000, hit_data);
		bfd_put_16 (input_bfd, 0xbf00, hit_data + 2);
	      }
	    return bfd_reloc_ok;
	  }
 
	/* Fetch the addend.  We use the Thumb-2 encoding (backwards compatible
           with Thumb-1) involving the J1 and J2 bits.  */
	if (globals->use_rel)
	  {
            bfd_vma s = (upper_insn & (1 << 10)) >> 10;
            bfd_vma upper = upper_insn & 0x3ff;
            bfd_vma lower = lower_insn & 0x7ff;
	    bfd_vma j1 = (lower_insn & (1 << 13)) >> 13;
	    bfd_vma j2 = (lower_insn & (1 << 11)) >> 11;
            bfd_vma i1 = j1 ^ s ? 0 : 1;
            bfd_vma i2 = j2 ^ s ? 0 : 1;
 
            addend = (i1 << 23) | (i2 << 22) | (upper << 12) | (lower << 1);
            /* Sign extend.  */
            addend = (addend | ((s ? 0 : 1) << 24)) - (1 << 24);
 
	    signed_addend = addend;
	  }
 
	if (r_type == R_ARM_THM_XPC22)
	  {
	    /* Check for Thumb to Thumb call.  */
	    /* FIXME: Should we translate the instruction into a BL
	       instruction instead ?  */
	    if (branch_type == ST_BRANCH_TO_THUMB)
	      (*_bfd_error_handler)
		(_("%B: Warning: Thumb BLX instruction targets thumb function '%s'."),
		 input_bfd,
		 h ? h->root.root.string : "(local)");
	  }
	else
	  {
	    /* If it is not a call to Thumb, assume call to Arm.
	       If it is a call relative to a section name, then it is not a
	       function call at all, but rather a long jump.  Calls through
	       the PLT do not require stubs.  */
	    if (branch_type == ST_BRANCH_TO_ARM && plt_offset == (bfd_vma) -1)
	      {
		if (globals->use_blx && r_type == R_ARM_THM_CALL)
		  {
		    /* Convert BL to BLX.  */
		    lower_insn = (lower_insn & ~0x1000) | 0x0800;
		  }
		else if ((   r_type != R_ARM_THM_CALL)
			 && (r_type != R_ARM_THM_JUMP24))
		  {
		    if (elf32_thumb_to_arm_stub
			(info, sym_name, input_bfd, output_bfd, input_section,
			 hit_data, sym_sec, rel->r_offset, signed_addend, value,
			 error_message))
		      return bfd_reloc_ok;
		    else
		      return bfd_reloc_dangerous;
		  }
	      }
	    else if (branch_type == ST_BRANCH_TO_THUMB
		     && globals->use_blx
		     && r_type == R_ARM_THM_CALL)
	      {
		/* Make sure this is a BL.  */
		lower_insn |= 0x1800;
	      }
	  }
 
	enum elf32_arm_stub_type stub_type = arm_stub_none;
	if (r_type == R_ARM_THM_CALL || r_type == R_ARM_THM_JUMP24)
	  {
	    /* Check if a stub has to be inserted because the destination
	       is too far.  */
	    struct elf32_arm_stub_hash_entry *stub_entry;
	    struct elf32_arm_link_hash_entry *hash;
 
	    hash = (struct elf32_arm_link_hash_entry *) h;
 
	    stub_type = arm_type_of_stub (info, input_section, rel,
					  st_type, &branch_type,
					  hash, value, sym_sec,
					  input_bfd, sym_name);
 
	    if (stub_type != arm_stub_none)
	      {
		/* The target is out of reach or we are changing modes, so
		   redirect the branch to the local stub for this
		   function.  */
		stub_entry = elf32_arm_get_stub_entry (input_section,
						       sym_sec, h,
						       rel, globals,
						       stub_type);
		if (stub_entry != NULL)
		  value = (stub_entry->stub_offset
			   + stub_entry->stub_sec->output_offset
			   + stub_entry->stub_sec->output_section->vma);
 
		/* If this call becomes a call to Arm, force BLX.  */
		if (globals->use_blx && (r_type == R_ARM_THM_CALL))
		  {
		    if ((stub_entry
			 && !arm_stub_is_thumb (stub_entry->stub_type))
			|| branch_type != ST_BRANCH_TO_THUMB)
		      lower_insn = (lower_insn & ~0x1000) | 0x0800;
		  }
	      }
	  }
 
	/* Handle calls via the PLT.  */
	if (stub_type == arm_stub_none && plt_offset != (bfd_vma) -1)
	  {
	    value = (splt->output_section->vma
		     + splt->output_offset
		     + plt_offset);
 
	    if (globals->use_blx && r_type == R_ARM_THM_CALL)
	      {
		/* If the Thumb BLX instruction is available, convert
		   the BL to a BLX instruction to call the ARM-mode
		   PLT entry.  */
		lower_insn = (lower_insn & ~0x1000) | 0x0800;
		branch_type = ST_BRANCH_TO_ARM;
	      }
	    else
	      {
		/* Target the Thumb stub before the ARM PLT entry.  */
		value -= PLT_THUMB_STUB_SIZE;
		branch_type = ST_BRANCH_TO_THUMB;
	      }
	    *unresolved_reloc_p = FALSE;
	  }
 
	relocation = value + signed_addend;
 
	relocation -= (input_section->output_section->vma
		       + input_section->output_offset
		       + rel->r_offset);
 
	check = relocation >> howto->rightshift;
 
	/* If this is a signed value, the rightshift just dropped
	   leading 1 bits (assuming twos complement).  */
	if ((bfd_signed_vma) relocation >= 0)
	  signed_check = check;
	else
	  signed_check = check | ~((bfd_vma) -1 >> howto->rightshift);
 
	/* Calculate the permissable maximum and minimum values for
	   this relocation according to whether we're relocating for
	   Thumb-2 or not.  */
	bitsize = howto->bitsize;
	if (!thumb2)
	  bitsize -= 2;
	reloc_signed_max = (1 << (bitsize - 1)) - 1;
	reloc_signed_min = ~reloc_signed_max;
 
	/* Assumes two's complement.  */
	if (signed_check > reloc_signed_max || signed_check < reloc_signed_min)
	  overflow = TRUE;
 
	if ((lower_insn & 0x5000) == 0x4000)
	  /* For a BLX instruction, make sure that the relocation is rounded up
	     to a word boundary.  This follows the semantics of the instruction
	     which specifies that bit 1 of the target address will come from bit
	     1 of the base address.  */
	  relocation = (relocation + 2) & ~ 3;
 
	/* Put RELOCATION back into the insn.  Assumes two's complement.
	   We use the Thumb-2 encoding, which is safe even if dealing with
	   a Thumb-1 instruction by virtue of our overflow check above.  */
        reloc_sign = (signed_check < 0) ? 1 : 0;
	upper_insn = (upper_insn & ~(bfd_vma) 0x7ff)
                     | ((relocation >> 12) & 0x3ff)
                     | (reloc_sign << 10);
	lower_insn = (lower_insn & ~(bfd_vma) 0x2fff)
                     | (((!((relocation >> 23) & 1)) ^ reloc_sign) << 13)
                     | (((!((relocation >> 22) & 1)) ^ reloc_sign) << 11)
                     | ((relocation >> 1) & 0x7ff);
 
	/* Put the relocated value back in the object file:  */
	bfd_put_16 (input_bfd, upper_insn, hit_data);
	bfd_put_16 (input_bfd, lower_insn, hit_data + 2);
 
	return (overflow ? bfd_reloc_overflow : bfd_reloc_ok);
      }
      break;
 
    case R_ARM_THM_JUMP19:
      /* Thumb32 conditional branch instruction.  */
      {
	bfd_vma relocation;
	bfd_boolean overflow = FALSE;
	bfd_vma upper_insn = bfd_get_16 (input_bfd, hit_data);
	bfd_vma lower_insn = bfd_get_16 (input_bfd, hit_data + 2);
	bfd_signed_vma reloc_signed_max = 0xffffe;
	bfd_signed_vma reloc_signed_min = -0x100000;
	bfd_signed_vma signed_check;
 
	/* Need to refetch the addend, reconstruct the top three bits,
	   and squish the two 11 bit pieces together.  */
	if (globals->use_rel)
	  {
	    bfd_vma S     = (upper_insn & 0x0400) >> 10;
	    bfd_vma upper = (upper_insn & 0x003f);
	    bfd_vma J1    = (lower_insn & 0x2000) >> 13;
	    bfd_vma J2    = (lower_insn & 0x0800) >> 11;
	    bfd_vma lower = (lower_insn & 0x07ff);
 
	    upper |= J1 << 6;
	    upper |= J2 << 7;
	    upper |= (!S) << 8;
	    upper -= 0x0100; /* Sign extend.  */
 
	    addend = (upper << 12) | (lower << 1);
	    signed_addend = addend;
	  }
 
	/* Handle calls via the PLT.  */
	if (plt_offset != (bfd_vma) -1)
	  {
	    value = (splt->output_section->vma
		     + splt->output_offset
		     + plt_offset);
	    /* Target the Thumb stub before the ARM PLT entry.  */
	    value -= PLT_THUMB_STUB_SIZE;
	    *unresolved_reloc_p = FALSE;
	  }
 
	/* ??? Should handle interworking?  GCC might someday try to
	   use this for tail calls.  */
 
      	relocation = value + signed_addend;
	relocation -= (input_section->output_section->vma
		       + input_section->output_offset
		       + rel->r_offset);
	signed_check = (bfd_signed_vma) relocation;
 
	if (signed_check > reloc_signed_max || signed_check < reloc_signed_min)
	  overflow = TRUE;
 
	/* Put RELOCATION back into the insn.  */
	{
	  bfd_vma S  = (relocation & 0x00100000) >> 20;
	  bfd_vma J2 = (relocation & 0x00080000) >> 19;
	  bfd_vma J1 = (relocation & 0x00040000) >> 18;
	  bfd_vma hi = (relocation & 0x0003f000) >> 12;
	  bfd_vma lo = (relocation & 0x00000ffe) >>  1;
 
	  upper_insn = (upper_insn & 0xfbc0) | (S << 10) | hi;
	  lower_insn = (lower_insn & 0xd000) | (J1 << 13) | (J2 << 11) | lo;
	}
 
	/* Put the relocated value back in the object file:  */
	bfd_put_16 (input_bfd, upper_insn, hit_data);
	bfd_put_16 (input_bfd, lower_insn, hit_data + 2);
 
	return (overflow ? bfd_reloc_overflow : bfd_reloc_ok);
      }
 
    case R_ARM_THM_JUMP11:
    case R_ARM_THM_JUMP8:
    case R_ARM_THM_JUMP6:
      /* Thumb B (branch) instruction).  */
      {
	bfd_signed_vma relocation;
	bfd_signed_vma reloc_signed_max = (1 << (howto->bitsize - 1)) - 1;
	bfd_signed_vma reloc_signed_min = ~ reloc_signed_max;
	bfd_signed_vma signed_check;
 
	/* CZB cannot jump backward.  */
	if (r_type == R_ARM_THM_JUMP6)
	  reloc_signed_min = 0;
 
	if (globals->use_rel)
	  {
	    /* Need to refetch addend.  */
	    addend = bfd_get_16 (input_bfd, hit_data) & howto->src_mask;
	    if (addend & ((howto->src_mask + 1) >> 1))
	      {
		signed_addend = -1;
		signed_addend &= ~ howto->src_mask;
		signed_addend |= addend;
	      }
	    else
	      signed_addend = addend;
	    /* The value in the insn has been right shifted.  We need to
	       undo this, so that we can perform the address calculation
	       in terms of bytes.  */
	    signed_addend <<= howto->rightshift;
	  }
	relocation = value + signed_addend;
 
	relocation -= (input_section->output_section->vma
		       + input_section->output_offset
		       + rel->r_offset);
 
	relocation >>= howto->rightshift;
	signed_check = relocation;
 
	if (r_type == R_ARM_THM_JUMP6)
	  relocation = ((relocation & 0x0020) << 4) | ((relocation & 0x001f) << 3);
	else
	  relocation &= howto->dst_mask;
	relocation |= (bfd_get_16 (input_bfd, hit_data) & (~ howto->dst_mask));
 
	bfd_put_16 (input_bfd, relocation, hit_data);
 
	/* Assumes two's complement.  */
	if (signed_check > reloc_signed_max || signed_check < reloc_signed_min)
	  return bfd_reloc_overflow;
 
	return bfd_reloc_ok;
      }
 
    case R_ARM_ALU_PCREL7_0:
    case R_ARM_ALU_PCREL15_8:
    case R_ARM_ALU_PCREL23_15:
      {
	bfd_vma insn;
	bfd_vma relocation;
 
	insn = bfd_get_32 (input_bfd, hit_data);
	if (globals->use_rel)
	  {
	    /* Extract the addend.  */
	    addend = (insn & 0xff) << ((insn & 0xf00) >> 7);
	    signed_addend = addend;
	  }
	relocation = value + signed_addend;
 
	relocation -= (input_section->output_section->vma
		       + input_section->output_offset
		       + rel->r_offset);
	insn = (insn & ~0xfff)
	       | ((howto->bitpos << 7) & 0xf00)
	       | ((relocation >> howto->bitpos) & 0xff);
	bfd_put_32 (input_bfd, value, hit_data);
      }
      return bfd_reloc_ok;
 
    case R_ARM_GNU_VTINHERIT:
    case R_ARM_GNU_VTENTRY:
      return bfd_reloc_ok;
 
    case R_ARM_GOTOFF32:
      /* Relocation is relative to the start of the
         global offset table.  */
 
      BFD_ASSERT (sgot != NULL);
      if (sgot == NULL)
        return bfd_reloc_notsupported;
 
      /* If we are addressing a Thumb function, we need to adjust the
	 address by one, so that attempts to call the function pointer will
	 correctly interpret it as Thumb code.  */
      if (branch_type == ST_BRANCH_TO_THUMB)
	value += 1;
 
      /* Note that sgot->output_offset is not involved in this
         calculation.  We always want the start of .got.  If we
         define _GLOBAL_OFFSET_TABLE in a different way, as is
         permitted by the ABI, we might have to change this
         calculation.  */
      value -= sgot->output_section->vma;
      return _bfd_final_link_relocate (howto, input_bfd, input_section,
				       contents, rel->r_offset, value,
				       rel->r_addend);
 
    case R_ARM_GOTPC:
      /* Use global offset table as symbol value.  */
      BFD_ASSERT (sgot != NULL);
 
      if (sgot == NULL)
        return bfd_reloc_notsupported;
 
      *unresolved_reloc_p = FALSE;
      value = sgot->output_section->vma;
      return _bfd_final_link_relocate (howto, input_bfd, input_section,
				       contents, rel->r_offset, value,
				       rel->r_addend);
 
    case R_ARM_GOT32:
    case R_ARM_GOT_PREL:
      /* Relocation is to the entry for this symbol in the
         global offset table.  */
      if (sgot == NULL)
	return bfd_reloc_notsupported;
 
      if (dynreloc_st_type == STT_GNU_IFUNC
	  && plt_offset != (bfd_vma) -1
	  && (h == NULL || SYMBOL_REFERENCES_LOCAL (info, h)))
	{
	  /* We have a relocation against a locally-binding STT_GNU_IFUNC
	     symbol, and the relocation resolves directly to the runtime
	     target rather than to the .iplt entry.  This means that any
	     .got entry would be the same value as the .igot.plt entry,
	     so there's no point creating both.  */
	  sgot = globals->root.igotplt;
	  value = sgot->output_offset + gotplt_offset;
	}
      else if (h != NULL)
	{
	  bfd_vma off;
 
	  off = h->got.offset;
	  BFD_ASSERT (off != (bfd_vma) -1);
	  if ((off & 1) != 0)
	    {
	      /* We have already processsed one GOT relocation against
		 this symbol.  */
	      off &= ~1;
	      if (globals->root.dynamic_sections_created
		  && !SYMBOL_REFERENCES_LOCAL (info, h))
		*unresolved_reloc_p = FALSE;
	    }
	  else
	    {
	      Elf_Internal_Rela outrel;
 
	      if (!SYMBOL_REFERENCES_LOCAL (info, h))
		{
		  /* If the symbol doesn't resolve locally in a static
		     object, we have an undefined reference.  If the
		     symbol doesn't resolve locally in a dynamic object,
		     it should be resolved by the dynamic linker.  */
		  if (globals->root.dynamic_sections_created)
		    {
		      outrel.r_info = ELF32_R_INFO (h->dynindx, R_ARM_GLOB_DAT);
		      *unresolved_reloc_p = FALSE;
		    }
		  else
		    outrel.r_info = 0;
		  outrel.r_addend = 0;
		}
	      else
		{
		  if (dynreloc_st_type == STT_GNU_IFUNC)
 		    outrel.r_info = ELF32_R_INFO (0, R_ARM_IRELATIVE);
		  else if (info->shared)
 		    outrel.r_info = ELF32_R_INFO (0, R_ARM_RELATIVE);
 		  else
 		    outrel.r_info = 0;
		  outrel.r_addend = dynreloc_value;
		}
 
	      /* The GOT entry is initialized to zero by default.
		 See if we should install a different value.  */
	      if (outrel.r_addend != 0
		  && (outrel.r_info == 0 || globals->use_rel))
		{
		  bfd_put_32 (output_bfd, outrel.r_addend,
			      sgot->contents + off);
		  outrel.r_addend = 0;
		}
 
	      if (outrel.r_info != 0)
		{
		  outrel.r_offset = (sgot->output_section->vma
				     + sgot->output_offset
				     + off);
		  elf32_arm_add_dynreloc (output_bfd, info, srelgot, &outrel);
		}
	      h->got.offset |= 1;
	    }
	  value = sgot->output_offset + off;
	}
      else
	{
	  bfd_vma off;
 
	  BFD_ASSERT (local_got_offsets != NULL &&
		      local_got_offsets[r_symndx] != (bfd_vma) -1);
 
	  off = local_got_offsets[r_symndx];
 
	  /* The offset must always be a multiple of 4.  We use the
	     least significant bit to record whether we have already
	     generated the necessary reloc.  */
	  if ((off & 1) != 0)
	    off &= ~1;
	  else
	    {
	      if (globals->use_rel)
		bfd_put_32 (output_bfd, dynreloc_value, sgot->contents + off);
 
	      if (info->shared || dynreloc_st_type == STT_GNU_IFUNC)
		{
		  Elf_Internal_Rela outrel;
 
		  outrel.r_addend = addend + dynreloc_value;
		  outrel.r_offset = (sgot->output_section->vma
				     + sgot->output_offset
				     + off);
		  if (dynreloc_st_type == STT_GNU_IFUNC)
 		    outrel.r_info = ELF32_R_INFO (0, R_ARM_IRELATIVE);
		  else
		    outrel.r_info = ELF32_R_INFO (0, R_ARM_RELATIVE);
		  elf32_arm_add_dynreloc (output_bfd, info, srelgot, &outrel);
		}
 
	      local_got_offsets[r_symndx] |= 1;
	    }
 
	  value = sgot->output_offset + off;
	}
      if (r_type != R_ARM_GOT32)
	value += sgot->output_section->vma;
 
      return _bfd_final_link_relocate (howto, input_bfd, input_section,
				       contents, rel->r_offset, value,
				       rel->r_addend);
 
    case R_ARM_TLS_LDO32:
      value = value - dtpoff_base (info);
 
      return _bfd_final_link_relocate (howto, input_bfd, input_section,
				       contents, rel->r_offset, value,
				       rel->r_addend);
 
    case R_ARM_TLS_LDM32:
      {
	bfd_vma off;
 
	if (sgot == NULL)
	  abort ();
 
	off = globals->tls_ldm_got.offset;
 
	if ((off & 1) != 0)
	  off &= ~1;
	else
	  {
	    /* If we don't know the module number, create a relocation
	       for it.  */
	    if (info->shared)
	      {
		Elf_Internal_Rela outrel;
 
		if (srelgot == NULL)
		  abort ();
 
		outrel.r_addend = 0;
		outrel.r_offset = (sgot->output_section->vma
				   + sgot->output_offset + off);
		outrel.r_info = ELF32_R_INFO (0, R_ARM_TLS_DTPMOD32);
 
		if (globals->use_rel)
		  bfd_put_32 (output_bfd, outrel.r_addend,
			      sgot->contents + off);
 
		elf32_arm_add_dynreloc (output_bfd, info, srelgot, &outrel);
	      }
	    else
	      bfd_put_32 (output_bfd, 1, sgot->contents + off);
 
	    globals->tls_ldm_got.offset |= 1;
	  }
 
	value = sgot->output_section->vma + sgot->output_offset + off
	  - (input_section->output_section->vma + input_section->output_offset + rel->r_offset);
 
	return _bfd_final_link_relocate (howto, input_bfd, input_section,
					 contents, rel->r_offset, value,
					 rel->r_addend);
      }
 
    case R_ARM_TLS_CALL:
    case R_ARM_THM_TLS_CALL:
    case R_ARM_TLS_GD32:
    case R_ARM_TLS_IE32:
    case R_ARM_TLS_GOTDESC:
    case R_ARM_TLS_DESCSEQ:
    case R_ARM_THM_TLS_DESCSEQ:
      {
	bfd_vma off, offplt;
	int indx = 0;
	char tls_type;
 
	BFD_ASSERT (sgot != NULL);
 
	if (h != NULL)
	  {
	    bfd_boolean dyn;
	    dyn = globals->root.dynamic_sections_created;
	    if (WILL_CALL_FINISH_DYNAMIC_SYMBOL (dyn, info->shared, h)
		&& (!info->shared
		    || !SYMBOL_REFERENCES_LOCAL (info, h)))
	      {
		*unresolved_reloc_p = FALSE;
		indx = h->dynindx;
	      }
	    off = h->got.offset;
	    offplt = elf32_arm_hash_entry (h)->tlsdesc_got;
	    tls_type = ((struct elf32_arm_link_hash_entry *) h)->tls_type;
	  }
	else
	  {
	    BFD_ASSERT (local_got_offsets != NULL);
	    off = local_got_offsets[r_symndx];
	    offplt = local_tlsdesc_gotents[r_symndx];
	    tls_type = elf32_arm_local_got_tls_type (input_bfd)[r_symndx];
	  }
 
	/* Linker relaxations happens from one of the
	   R_ARM_{GOTDESC,CALL,DESCSEQ} relocations to IE or LE.  */ 
	if (ELF32_R_TYPE(rel->r_info) != r_type)
	  tls_type = GOT_TLS_IE; 
 
	BFD_ASSERT (tls_type != GOT_UNKNOWN);
 
	if ((off & 1) != 0)
	  off &= ~1;
	else
	  {
	    bfd_boolean need_relocs = FALSE;
	    Elf_Internal_Rela outrel;
	    int cur_off = off;
 
	    /* The GOT entries have not been initialized yet.  Do it
	       now, and emit any relocations.  If both an IE GOT and a
	       GD GOT are necessary, we emit the GD first.  */
 
	    if ((info->shared || indx != 0)
		&& (h == NULL
		    || ELF_ST_VISIBILITY (h->other) == STV_DEFAULT
		    || h->root.type != bfd_link_hash_undefweak))
	      {
		need_relocs = TRUE;
		BFD_ASSERT (srelgot != NULL);
	      }
 
	    if (tls_type & GOT_TLS_GDESC)
	      {
		bfd_byte *loc;
 
		/* We should have relaxed, unless this is an undefined
		   weak symbol.  */
		BFD_ASSERT ((h && (h->root.type == bfd_link_hash_undefweak))
			    || info->shared);
		BFD_ASSERT (globals->sgotplt_jump_table_size + offplt + 8
                            <= globals->root.sgotplt->size);
 
		outrel.r_addend = 0;
		outrel.r_offset = (globals->root.sgotplt->output_section->vma
				   + globals->root.sgotplt->output_offset
				   + offplt
				   + globals->sgotplt_jump_table_size);
 
		outrel.r_info = ELF32_R_INFO (indx, R_ARM_TLS_DESC);
		sreloc = globals->root.srelplt;
		loc = sreloc->contents;
		loc += globals->next_tls_desc_index++ * RELOC_SIZE (globals);
		BFD_ASSERT (loc + RELOC_SIZE (globals)
		   	   <= sreloc->contents + sreloc->size);
 
		SWAP_RELOC_OUT (globals) (output_bfd, &outrel, loc);
 
		/* For globals, the first word in the relocation gets
		   the relocation index and the top bit set, or zero,
		   if we're binding now.  For locals, it gets the
		   symbol's offset in the tls section.  */
	    	bfd_put_32 (output_bfd,
			    !h ? value - elf_hash_table (info)->tls_sec->vma
			    : info->flags & DF_BIND_NOW ? 0
			    : 0x80000000 | ELF32_R_SYM (outrel.r_info),
			    globals->root.sgotplt->contents + offplt +
			    globals->sgotplt_jump_table_size);
 
		/* Second word in the relocation is always zero.  */
	    	bfd_put_32 (output_bfd, 0,
			    globals->root.sgotplt->contents + offplt +
			    globals->sgotplt_jump_table_size + 4);
	      }
	    if (tls_type & GOT_TLS_GD)
	      {
		if (need_relocs)
		  {
		    outrel.r_addend = 0;
		    outrel.r_offset = (sgot->output_section->vma
				       + sgot->output_offset
				       + cur_off);
		    outrel.r_info = ELF32_R_INFO (indx, R_ARM_TLS_DTPMOD32);
 
		    if (globals->use_rel)
		      bfd_put_32 (output_bfd, outrel.r_addend,
				  sgot->contents + cur_off);
 
		    elf32_arm_add_dynreloc (output_bfd, info, srelgot, &outrel);
 
		    if (indx == 0)
		      bfd_put_32 (output_bfd, value - dtpoff_base (info),
				  sgot->contents + cur_off + 4);
		    else
		      {
			outrel.r_addend = 0;
			outrel.r_info = ELF32_R_INFO (indx,
						      R_ARM_TLS_DTPOFF32);
			outrel.r_offset += 4;
 
			if (globals->use_rel)
			  bfd_put_32 (output_bfd, outrel.r_addend,
				      sgot->contents + cur_off + 4);
 
			elf32_arm_add_dynreloc (output_bfd, info,
						srelgot, &outrel);
		      }
		  }
		else
		  {
		    /* If we are not emitting relocations for a
		       general dynamic reference, then we must be in a
		       static link or an executable link with the
		       symbol binding locally.  Mark it as belonging
		       to module 1, the executable.  */
		    bfd_put_32 (output_bfd, 1,
				sgot->contents + cur_off);
		    bfd_put_32 (output_bfd, value - dtpoff_base (info),
				sgot->contents + cur_off + 4);
		  }
 
		cur_off += 8;
	      }
 
	    if (tls_type & GOT_TLS_IE)
	      {
		if (need_relocs)
		  {
		    if (indx == 0)
		      outrel.r_addend = value - dtpoff_base (info);
		    else
		      outrel.r_addend = 0;
		    outrel.r_offset = (sgot->output_section->vma
				       + sgot->output_offset
				       + cur_off);
		    outrel.r_info = ELF32_R_INFO (indx, R_ARM_TLS_TPOFF32);
 
		    if (globals->use_rel)
		      bfd_put_32 (output_bfd, outrel.r_addend,
				  sgot->contents + cur_off);
 
		    elf32_arm_add_dynreloc (output_bfd, info, srelgot, &outrel);
		  }
		else
		  bfd_put_32 (output_bfd, tpoff (info, value),
			      sgot->contents + cur_off);
		cur_off += 4;
	      }
 
	    if (h != NULL)
	      h->got.offset |= 1;
	    else
	      local_got_offsets[r_symndx] |= 1;
	  }
 
	if ((tls_type & GOT_TLS_GD) && r_type != R_ARM_TLS_GD32)
	  off += 8;
	else if (tls_type & GOT_TLS_GDESC)
	  off = offplt;
 
	if (ELF32_R_TYPE(rel->r_info) == R_ARM_TLS_CALL
	    || ELF32_R_TYPE(rel->r_info) == R_ARM_THM_TLS_CALL)
	  {
	    bfd_signed_vma offset;
	    /* TLS stubs are arm mode.  The original symbol is a
	       data object, so branch_type is bogus.  */
	    branch_type = ST_BRANCH_TO_ARM;
	    enum elf32_arm_stub_type stub_type
	      = arm_type_of_stub (info, input_section, rel,
				  st_type, &branch_type,
				  (struct elf32_arm_link_hash_entry *)h,
				  globals->tls_trampoline, globals->root.splt,
				  input_bfd, sym_name);
 
	    if (stub_type != arm_stub_none)
	      {
		struct elf32_arm_stub_hash_entry *stub_entry
		  = elf32_arm_get_stub_entry
		  (input_section, globals->root.splt, 0, rel,
		   globals, stub_type);
		offset = (stub_entry->stub_offset
			  + stub_entry->stub_sec->output_offset
			  + stub_entry->stub_sec->output_section->vma);
	      }
	    else
	      offset = (globals->root.splt->output_section->vma
			+ globals->root.splt->output_offset
			+ globals->tls_trampoline);
 
	    if (ELF32_R_TYPE(rel->r_info) == R_ARM_TLS_CALL)
	      {
		unsigned long inst;
 
		offset -= (input_section->output_section->vma +
			   input_section->output_offset + rel->r_offset + 8);
 
		inst = offset >> 2;
		inst &= 0x00ffffff;
		value = inst | (globals->use_blx ? 0xfa000000 : 0xeb000000);
	      }
	    else
	      {
		/* Thumb blx encodes the offset in a complicated
		   fashion.  */
		unsigned upper_insn, lower_insn;
		unsigned neg;
 
		offset -= (input_section->output_section->vma + 
			   input_section->output_offset
			   + rel->r_offset + 4);
 
		if (stub_type != arm_stub_none
		    && arm_stub_is_thumb (stub_type))
		  {
		    lower_insn = 0xd000;
		  }
		else
		  {
		    lower_insn = 0xc000;
		    /* Round up the offset to a word boundary */
		    offset = (offset + 2) & ~2;
		  }
 
		neg = offset < 0;
		upper_insn = (0xf000
			      | ((offset >> 12) & 0x3ff)
			      | (neg << 10));
		lower_insn |= (((!((offset >> 23) & 1)) ^ neg) << 13)
			      | (((!((offset >> 22) & 1)) ^ neg) << 11)
			      | ((offset >> 1) & 0x7ff);
		bfd_put_16 (input_bfd, upper_insn, hit_data);
		bfd_put_16 (input_bfd, lower_insn, hit_data + 2);
		return bfd_reloc_ok;
	      }
	  }
	/* These relocations needs special care, as besides the fact
	   they point somewhere in .gotplt, the addend must be
	   adjusted accordingly depending on the type of instruction
	   we refer to */
	else if ((r_type == R_ARM_TLS_GOTDESC) && (tls_type & GOT_TLS_GDESC))
	  {
	    unsigned long data, insn;
	    unsigned thumb;
 
	    data = bfd_get_32 (input_bfd, hit_data);
	    thumb = data & 1;
	    data &= ~1u;
 
	    if (thumb)
	      {
		insn = bfd_get_16 (input_bfd, contents + rel->r_offset - data);
		if ((insn & 0xf000) == 0xf000 || (insn & 0xf800) == 0xe800)
		  insn = (insn << 16)
		    | bfd_get_16 (input_bfd,
				  contents + rel->r_offset - data + 2);
		if ((insn & 0xf800c000) == 0xf000c000)
		  /* bl/blx */
		  value = -6;
		else if ((insn & 0xffffff00) == 0x4400)
		  /* add */
		  value = -5;
		else
		  {
		    (*_bfd_error_handler)
		      (_("%B(%A+0x%lx):unexpected Thumb instruction '0x%x' referenced by TLS_GOTDESC"),
		       input_bfd, input_section,
		       (unsigned long)rel->r_offset, insn);
		    return bfd_reloc_notsupported;
		  }
	      }
	    else
	      {
		insn = bfd_get_32 (input_bfd, contents + rel->r_offset - data);
 
		switch (insn >> 24)
		  {
		  case 0xeb:  /* bl */
		  case 0xfa:  /* blx */
		    value = -4;
		    break;
 
		  case 0xe0:	/* add */
		    value = -8;
		    break;
 
		  default:
		    (*_bfd_error_handler)
		      (_("%B(%A+0x%lx):unexpected ARM instruction '0x%x' referenced by TLS_GOTDESC"),
		       input_bfd, input_section,
		       (unsigned long)rel->r_offset, insn);
		    return bfd_reloc_notsupported;
		  }
	      }
 
	    value += ((globals->root.sgotplt->output_section->vma
		       + globals->root.sgotplt->output_offset + off)
		      - (input_section->output_section->vma
			 + input_section->output_offset
			 + rel->r_offset)
		      + globals->sgotplt_jump_table_size);
	  }
	else
	  value = ((globals->root.sgot->output_section->vma
		    + globals->root.sgot->output_offset + off)
		   - (input_section->output_section->vma
		      + input_section->output_offset + rel->r_offset));
 
	return _bfd_final_link_relocate (howto, input_bfd, input_section,
					 contents, rel->r_offset, value,
					 rel->r_addend);
      }
 
    case R_ARM_TLS_LE32:
      if (info->shared)
	{
	  (*_bfd_error_handler)
	    (_("%B(%A+0x%lx): R_ARM_TLS_LE32 relocation not permitted in shared object"),
	     input_bfd, input_section,
	     (long) rel->r_offset, howto->name);
	  return (bfd_reloc_status_type) FALSE;
	}
      else
	value = tpoff (info, value);
 
      return _bfd_final_link_relocate (howto, input_bfd, input_section,
				       contents, rel->r_offset, value,
				       rel->r_addend);
 
    case R_ARM_V4BX:
      if (globals->fix_v4bx)
	{
	  bfd_vma insn = bfd_get_32 (input_bfd, hit_data);
 
	  /* Ensure that we have a BX instruction.  */
	  BFD_ASSERT ((insn & 0x0ffffff0) == 0x012fff10);
 
	  if (globals->fix_v4bx == 2 && (insn & 0xf) != 0xf)
	    {
	      /* Branch to veneer.  */
	      bfd_vma glue_addr;
	      glue_addr = elf32_arm_bx_glue (info, insn & 0xf);
	      glue_addr -= input_section->output_section->vma
			   + input_section->output_offset
			   + rel->r_offset + 8;
	      insn = (insn & 0xf0000000) | 0x0a000000
		     | ((glue_addr >> 2) & 0x00ffffff);
	    }
	  else
	    {
	      /* Preserve Rm (lowest four bits) and the condition code
		 (highest four bits). Other bits encode MOV PC,Rm.  */
	      insn = (insn & 0xf000000f) | 0x01a0f000;
	    }
 
	  bfd_put_32 (input_bfd, insn, hit_data);
	}
      return bfd_reloc_ok;
 
    case R_ARM_MOVW_ABS_NC:
    case R_ARM_MOVT_ABS:
    case R_ARM_MOVW_PREL_NC:
    case R_ARM_MOVT_PREL:
    /* Until we properly support segment-base-relative addressing then
       we assume the segment base to be zero, as for the group relocations.
       Thus R_ARM_MOVW_BREL_NC has the same semantics as R_ARM_MOVW_ABS_NC
       and R_ARM_MOVT_BREL has the same semantics as R_ARM_MOVT_ABS.  */
    case R_ARM_MOVW_BREL_NC:
    case R_ARM_MOVW_BREL:
    case R_ARM_MOVT_BREL:
      {
	bfd_vma insn = bfd_get_32 (input_bfd, hit_data);
 
	if (globals->use_rel)
	  {
	    addend = ((insn >> 4) & 0xf000) | (insn & 0xfff);
	    signed_addend = (addend ^ 0x8000) - 0x8000;
	  }
 
	value += signed_addend;
 
	if (r_type == R_ARM_MOVW_PREL_NC || r_type == R_ARM_MOVT_PREL)
	  value -= (input_section->output_section->vma
		    + input_section->output_offset + rel->r_offset);
 
	if (r_type == R_ARM_MOVW_BREL && value >= 0x10000)
          return bfd_reloc_overflow;
 
	if (branch_type == ST_BRANCH_TO_THUMB)
	  value |= 1;
 
	if (r_type == R_ARM_MOVT_ABS || r_type == R_ARM_MOVT_PREL
            || r_type == R_ARM_MOVT_BREL)
	  value >>= 16;
 
	insn &= 0xfff0f000;
	insn |= value & 0xfff;
	insn |= (value & 0xf000) << 4;
	bfd_put_32 (input_bfd, insn, hit_data);
      }
      return bfd_reloc_ok;
 
    case R_ARM_THM_MOVW_ABS_NC:
    case R_ARM_THM_MOVT_ABS:
    case R_ARM_THM_MOVW_PREL_NC:
    case R_ARM_THM_MOVT_PREL:
    /* Until we properly support segment-base-relative addressing then
       we assume the segment base to be zero, as for the above relocations.
       Thus R_ARM_THM_MOVW_BREL_NC has the same semantics as
       R_ARM_THM_MOVW_ABS_NC and R_ARM_THM_MOVT_BREL has the same semantics
       as R_ARM_THM_MOVT_ABS.  */
    case R_ARM_THM_MOVW_BREL_NC:
    case R_ARM_THM_MOVW_BREL:
    case R_ARM_THM_MOVT_BREL:
      {
	bfd_vma insn;
 
	insn = bfd_get_16 (input_bfd, hit_data) << 16;
	insn |= bfd_get_16 (input_bfd, hit_data + 2);
 
	if (globals->use_rel)
	  {
	    addend = ((insn >> 4)  & 0xf000)
		   | ((insn >> 15) & 0x0800)
		   | ((insn >> 4)  & 0x0700)
		   | (insn         & 0x00ff);
	    signed_addend = (addend ^ 0x8000) - 0x8000;
	  }
 
	value += signed_addend;
 
	if (r_type == R_ARM_THM_MOVW_PREL_NC || r_type == R_ARM_THM_MOVT_PREL)
	  value -= (input_section->output_section->vma
		    + input_section->output_offset + rel->r_offset);
 
	if (r_type == R_ARM_THM_MOVW_BREL && value >= 0x10000)
          return bfd_reloc_overflow;
 
	if (branch_type == ST_BRANCH_TO_THUMB)
	  value |= 1;
 
	if (r_type == R_ARM_THM_MOVT_ABS || r_type == R_ARM_THM_MOVT_PREL
            || r_type == R_ARM_THM_MOVT_BREL)
	  value >>= 16;
 
	insn &= 0xfbf08f00;
	insn |= (value & 0xf000) << 4;
	insn |= (value & 0x0800) << 15;
	insn |= (value & 0x0700) << 4;
	insn |= (value & 0x00ff);
 
	bfd_put_16 (input_bfd, insn >> 16, hit_data);
	bfd_put_16 (input_bfd, insn & 0xffff, hit_data + 2);
      }
      return bfd_reloc_ok;
 
    case R_ARM_ALU_PC_G0_NC:
    case R_ARM_ALU_PC_G1_NC:
    case R_ARM_ALU_PC_G0:
    case R_ARM_ALU_PC_G1:
    case R_ARM_ALU_PC_G2:
    case R_ARM_ALU_SB_G0_NC:
    case R_ARM_ALU_SB_G1_NC:
    case R_ARM_ALU_SB_G0:
    case R_ARM_ALU_SB_G1:
    case R_ARM_ALU_SB_G2:
      {
	bfd_vma insn = bfd_get_32 (input_bfd, hit_data);
        bfd_vma pc = input_section->output_section->vma
		     + input_section->output_offset + rel->r_offset;
        /* sb should be the origin of the *segment* containing the symbol.
           It is not clear how to obtain this OS-dependent value, so we
           make an arbitrary choice of zero.  */
        bfd_vma sb = 0;
        bfd_vma residual;
        bfd_vma g_n;
	bfd_signed_vma signed_value;
        int group = 0;
 
        /* Determine which group of bits to select.  */
        switch (r_type)
          {
          case R_ARM_ALU_PC_G0_NC:
          case R_ARM_ALU_PC_G0:
          case R_ARM_ALU_SB_G0_NC:
          case R_ARM_ALU_SB_G0:
            group = 0;
            break;
 
          case R_ARM_ALU_PC_G1_NC:
          case R_ARM_ALU_PC_G1:
          case R_ARM_ALU_SB_G1_NC:
          case R_ARM_ALU_SB_G1:
            group = 1;
            break;
 
          case R_ARM_ALU_PC_G2:
          case R_ARM_ALU_SB_G2:
            group = 2;
            break;
 
          default:
            abort ();
          }
 
        /* If REL, extract the addend from the insn.  If RELA, it will
           have already been fetched for us.  */
	if (globals->use_rel)
          {
            int negative;
            bfd_vma constant = insn & 0xff;
            bfd_vma rotation = (insn & 0xf00) >> 8;
 
            if (rotation == 0)
              signed_addend = constant;
            else
              {
                /* Compensate for the fact that in the instruction, the
                   rotation is stored in multiples of 2 bits.  */
                rotation *= 2;
 
                /* Rotate "constant" right by "rotation" bits.  */
                signed_addend = (constant >> rotation) |
                                (constant << (8 * sizeof (bfd_vma) - rotation));
              }
 
            /* Determine if the instruction is an ADD or a SUB.
               (For REL, this determines the sign of the addend.)  */
            negative = identify_add_or_sub (insn);
            if (negative == 0)
              {
                (*_bfd_error_handler)
                  (_("%B(%A+0x%lx): Only ADD or SUB instructions are allowed for ALU group relocations"),
                  input_bfd, input_section,
                  (long) rel->r_offset, howto->name);
                return bfd_reloc_overflow;
    	      }
 
            signed_addend *= negative;
          }
 
	/* Compute the value (X) to go in the place.  */
        if (r_type == R_ARM_ALU_PC_G0_NC
            || r_type == R_ARM_ALU_PC_G1_NC
            || r_type == R_ARM_ALU_PC_G0
            || r_type == R_ARM_ALU_PC_G1
            || r_type == R_ARM_ALU_PC_G2)
          /* PC relative.  */
          signed_value = value - pc + signed_addend;
        else
          /* Section base relative.  */
          signed_value = value - sb + signed_addend;
 
        /* If the target symbol is a Thumb function, then set the
           Thumb bit in the address.  */
	if (branch_type == ST_BRANCH_TO_THUMB)
	  signed_value |= 1;
 
        /* Calculate the value of the relevant G_n, in encoded
           constant-with-rotation format.  */
        g_n = calculate_group_reloc_mask (abs (signed_value), group,
                                          &residual);
 
        /* Check for overflow if required.  */
        if ((r_type == R_ARM_ALU_PC_G0
             || r_type == R_ARM_ALU_PC_G1
             || r_type == R_ARM_ALU_PC_G2
             || r_type == R_ARM_ALU_SB_G0
             || r_type == R_ARM_ALU_SB_G1
             || r_type == R_ARM_ALU_SB_G2) && residual != 0)
          {
            (*_bfd_error_handler)
              (_("%B(%A+0x%lx): Overflow whilst splitting 0x%lx for group relocation %s"),
              input_bfd, input_section,
              (long) rel->r_offset, abs (signed_value), howto->name);
            return bfd_reloc_overflow;
          }
 
        /* Mask out the value and the ADD/SUB part of the opcode; take care
           not to destroy the S bit.  */
        insn &= 0xff1ff000;
 
        /* Set the opcode according to whether the value to go in the
           place is negative.  */
        if (signed_value < 0)
          insn |= 1 << 22;
        else
          insn |= 1 << 23;
 
        /* Encode the offset.  */
        insn |= g_n;
 
	bfd_put_32 (input_bfd, insn, hit_data);
      }
      return bfd_reloc_ok;
 
    case R_ARM_LDR_PC_G0:
    case R_ARM_LDR_PC_G1:
    case R_ARM_LDR_PC_G2:
    case R_ARM_LDR_SB_G0:
    case R_ARM_LDR_SB_G1:
    case R_ARM_LDR_SB_G2:
      {
	bfd_vma insn = bfd_get_32 (input_bfd, hit_data);
        bfd_vma pc = input_section->output_section->vma
		     + input_section->output_offset + rel->r_offset;
        bfd_vma sb = 0; /* See note above.  */
        bfd_vma residual;
	bfd_signed_vma signed_value;
        int group = 0;
 
        /* Determine which groups of bits to calculate.  */
        switch (r_type)
          {
          case R_ARM_LDR_PC_G0:
          case R_ARM_LDR_SB_G0:
            group = 0;
            break;
 
          case R_ARM_LDR_PC_G1:
          case R_ARM_LDR_SB_G1:
            group = 1;
            break;
 
          case R_ARM_LDR_PC_G2:
          case R_ARM_LDR_SB_G2:
            group = 2;
            break;
 
          default:
            abort ();
          }
 
        /* If REL, extract the addend from the insn.  If RELA, it will
           have already been fetched for us.  */
	if (globals->use_rel)
          {
            int negative = (insn & (1 << 23)) ? 1 : -1;
            signed_addend = negative * (insn & 0xfff);
          }
 
	/* Compute the value (X) to go in the place.  */
        if (r_type == R_ARM_LDR_PC_G0
            || r_type == R_ARM_LDR_PC_G1
            || r_type == R_ARM_LDR_PC_G2)
          /* PC relative.  */
          signed_value = value - pc + signed_addend;
        else
          /* Section base relative.  */
          signed_value = value - sb + signed_addend;
 
        /* Calculate the value of the relevant G_{n-1} to obtain
           the residual at that stage.  */
        calculate_group_reloc_mask (abs (signed_value), group - 1, &residual);
 
        /* Check for overflow.  */
        if (residual >= 0x1000)
          {
            (*_bfd_error_handler)
              (_("%B(%A+0x%lx): Overflow whilst splitting 0x%lx for group relocation %s"),
              input_bfd, input_section,
              (long) rel->r_offset, abs (signed_value), howto->name);
            return bfd_reloc_overflow;
          }
 
        /* Mask out the value and U bit.  */
        insn &= 0xff7ff000;
 
        /* Set the U bit if the value to go in the place is non-negative.  */
        if (signed_value >= 0)
          insn |= 1 << 23;
 
        /* Encode the offset.  */
        insn |= residual;
 
	bfd_put_32 (input_bfd, insn, hit_data);
      }
      return bfd_reloc_ok;
 
    case R_ARM_LDRS_PC_G0:
    case R_ARM_LDRS_PC_G1:
    case R_ARM_LDRS_PC_G2:
    case R_ARM_LDRS_SB_G0:
    case R_ARM_LDRS_SB_G1:
    case R_ARM_LDRS_SB_G2:
      {
	bfd_vma insn = bfd_get_32 (input_bfd, hit_data);
        bfd_vma pc = input_section->output_section->vma
		     + input_section->output_offset + rel->r_offset;
        bfd_vma sb = 0; /* See note above.  */
        bfd_vma residual;
	bfd_signed_vma signed_value;
        int group = 0;
 
        /* Determine which groups of bits to calculate.  */
        switch (r_type)
          {
          case R_ARM_LDRS_PC_G0:
          case R_ARM_LDRS_SB_G0:
            group = 0;
            break;
 
          case R_ARM_LDRS_PC_G1:
          case R_ARM_LDRS_SB_G1:
            group = 1;
            break;
 
          case R_ARM_LDRS_PC_G2:
          case R_ARM_LDRS_SB_G2:
            group = 2;
            break;
 
          default:
            abort ();
          }
 
        /* If REL, extract the addend from the insn.  If RELA, it will
           have already been fetched for us.  */
	if (globals->use_rel)
          {
            int negative = (insn & (1 << 23)) ? 1 : -1;
            signed_addend = negative * (((insn & 0xf00) >> 4) + (insn & 0xf));
          }
 
	/* Compute the value (X) to go in the place.  */
        if (r_type == R_ARM_LDRS_PC_G0
            || r_type == R_ARM_LDRS_PC_G1
            || r_type == R_ARM_LDRS_PC_G2)
          /* PC relative.  */
          signed_value = value - pc + signed_addend;
        else
          /* Section base relative.  */
          signed_value = value - sb + signed_addend;
 
        /* Calculate the value of the relevant G_{n-1} to obtain
           the residual at that stage.  */
        calculate_group_reloc_mask (abs (signed_value), group - 1, &residual);
 
        /* Check for overflow.  */
        if (residual >= 0x100)
          {
            (*_bfd_error_handler)
              (_("%B(%A+0x%lx): Overflow whilst splitting 0x%lx for group relocation %s"),
              input_bfd, input_section,
              (long) rel->r_offset, abs (signed_value), howto->name);
            return bfd_reloc_overflow;
          }
 
        /* Mask out the value and U bit.  */
        insn &= 0xff7ff0f0;
 
        /* Set the U bit if the value to go in the place is non-negative.  */
        if (signed_value >= 0)
          insn |= 1 << 23;
 
        /* Encode the offset.  */
        insn |= ((residual & 0xf0) << 4) | (residual & 0xf);
 
	bfd_put_32 (input_bfd, insn, hit_data);
      }
      return bfd_reloc_ok;
 
    case R_ARM_LDC_PC_G0:
    case R_ARM_LDC_PC_G1:
    case R_ARM_LDC_PC_G2:
    case R_ARM_LDC_SB_G0:
    case R_ARM_LDC_SB_G1:
    case R_ARM_LDC_SB_G2:
      {
	bfd_vma insn = bfd_get_32 (input_bfd, hit_data);
        bfd_vma pc = input_section->output_section->vma
		     + input_section->output_offset + rel->r_offset;
        bfd_vma sb = 0; /* See note above.  */
        bfd_vma residual;
	bfd_signed_vma signed_value;
        int group = 0;
 
        /* Determine which groups of bits to calculate.  */
        switch (r_type)
          {
          case R_ARM_LDC_PC_G0:
          case R_ARM_LDC_SB_G0:
            group = 0;
            break;
 
          case R_ARM_LDC_PC_G1:
          case R_ARM_LDC_SB_G1:
            group = 1;
            break;
 
          case R_ARM_LDC_PC_G2:
          case R_ARM_LDC_SB_G2:
            group = 2;
            break;
 
          default:
            abort ();
          }
 
        /* If REL, extract the addend from the insn.  If RELA, it will
           have already been fetched for us.  */
	if (globals->use_rel)
          {
            int negative = (insn & (1 << 23)) ? 1 : -1;
            signed_addend = negative * ((insn & 0xff) << 2);
          }
 
	/* Compute the value (X) to go in the place.  */
        if (r_type == R_ARM_LDC_PC_G0
            || r_type == R_ARM_LDC_PC_G1
            || r_type == R_ARM_LDC_PC_G2)
          /* PC relative.  */
          signed_value = value - pc + signed_addend;
        else
          /* Section base relative.  */
          signed_value = value - sb + signed_addend;
 
        /* Calculate the value of the relevant G_{n-1} to obtain
           the residual at that stage.  */
        calculate_group_reloc_mask (abs (signed_value), group - 1, &residual);
 
        /* Check for overflow.  (The absolute value to go in the place must be
           divisible by four and, after having been divided by four, must
           fit in eight bits.)  */
        if ((residual & 0x3) != 0 || residual >= 0x400)
          {
            (*_bfd_error_handler)
              (_("%B(%A+0x%lx): Overflow whilst splitting 0x%lx for group relocation %s"),
              input_bfd, input_section,
              (long) rel->r_offset, abs (signed_value), howto->name);
            return bfd_reloc_overflow;
          }
 
        /* Mask out the value and U bit.  */
        insn &= 0xff7fff00;
 
        /* Set the U bit if the value to go in the place is non-negative.  */
        if (signed_value >= 0)
          insn |= 1 << 23;
 
        /* Encode the offset.  */
        insn |= residual >> 2;
 
	bfd_put_32 (input_bfd, insn, hit_data);
      }
      return bfd_reloc_ok;
 
    default:
      return bfd_reloc_notsupported;
    }
}
 
/* Add INCREMENT to the reloc (of type HOWTO) at ADDRESS.  */
static void
arm_add_to_rel (bfd *              abfd,
		bfd_byte *         address,
		reloc_howto_type * howto,
		bfd_signed_vma     increment)
{
  bfd_signed_vma addend;
 
  if (howto->type == R_ARM_THM_CALL
      || howto->type == R_ARM_THM_JUMP24)
    {
      int upper_insn, lower_insn;
      int upper, lower;
 
      upper_insn = bfd_get_16 (abfd, address);
      lower_insn = bfd_get_16 (abfd, address + 2);
      upper = upper_insn & 0x7ff;
      lower = lower_insn & 0x7ff;
 
      addend = (upper << 12) | (lower << 1);
      addend += increment;
      addend >>= 1;
 
      upper_insn = (upper_insn & 0xf800) | ((addend >> 11) & 0x7ff);
      lower_insn = (lower_insn & 0xf800) | (addend & 0x7ff);
 
      bfd_put_16 (abfd, (bfd_vma) upper_insn, address);
      bfd_put_16 (abfd, (bfd_vma) lower_insn, address + 2);
    }
  else
    {
      bfd_vma        contents;
 
      contents = bfd_get_32 (abfd, address);
 
      /* Get the (signed) value from the instruction.  */
      addend = contents & howto->src_mask;
      if (addend & ((howto->src_mask + 1) >> 1))
	{
	  bfd_signed_vma mask;
 
	  mask = -1;
	  mask &= ~ howto->src_mask;
	  addend |= mask;
	}
 
      /* Add in the increment, (which is a byte value).  */
      switch (howto->type)
	{
	default:
	  addend += increment;
	  break;
 
	case R_ARM_PC24:
	case R_ARM_PLT32:
	case R_ARM_CALL:
	case R_ARM_JUMP24:
	  addend <<= howto->size;
	  addend += increment;
 
	  /* Should we check for overflow here ?  */
 
	  /* Drop any undesired bits.  */
	  addend >>= howto->rightshift;
	  break;
	}
 
      contents = (contents & ~ howto->dst_mask) | (addend & howto->dst_mask);
 
      bfd_put_32 (abfd, contents, address);
    }
}
 
#define IS_ARM_TLS_RELOC(R_TYPE)	\
  ((R_TYPE) == R_ARM_TLS_GD32		\
   || (R_TYPE) == R_ARM_TLS_LDO32	\
   || (R_TYPE) == R_ARM_TLS_LDM32	\
   || (R_TYPE) == R_ARM_TLS_DTPOFF32	\
   || (R_TYPE) == R_ARM_TLS_DTPMOD32	\
   || (R_TYPE) == R_ARM_TLS_TPOFF32	\
   || (R_TYPE) == R_ARM_TLS_LE32	\
   || (R_TYPE) == R_ARM_TLS_IE32	\
   || IS_ARM_TLS_GNU_RELOC (R_TYPE))
 
/* Specific set of relocations for the gnu tls dialect.  */
#define IS_ARM_TLS_GNU_RELOC(R_TYPE)	\
  ((R_TYPE) == R_ARM_TLS_GOTDESC	\
   || (R_TYPE) == R_ARM_TLS_CALL	\
   || (R_TYPE) == R_ARM_THM_TLS_CALL	\
   || (R_TYPE) == R_ARM_TLS_DESCSEQ	\
   || (R_TYPE) == R_ARM_THM_TLS_DESCSEQ)
 
/* Relocate an ARM ELF section.  */
 
static bfd_boolean
elf32_arm_relocate_section (bfd *                  output_bfd,
			    struct bfd_link_info * info,
			    bfd *                  input_bfd,
			    asection *             input_section,
			    bfd_byte *             contents,
			    Elf_Internal_Rela *    relocs,
			    Elf_Internal_Sym *     local_syms,
			    asection **            local_sections)
{
  Elf_Internal_Shdr *symtab_hdr;
  struct elf_link_hash_entry **sym_hashes;
  Elf_Internal_Rela *rel;
  Elf_Internal_Rela *relend;
  const char *name;
  struct elf32_arm_link_hash_table * globals;
 
  globals = elf32_arm_hash_table (info);
  if (globals == NULL)
    return FALSE;
 
  symtab_hdr = & elf_symtab_hdr (input_bfd);
  sym_hashes = elf_sym_hashes (input_bfd);
 
  rel = relocs;
  relend = relocs + input_section->reloc_count;
  for (; rel < relend; rel++)
    {
      int                          r_type;
      reloc_howto_type *           howto;
      unsigned long                r_symndx;
      Elf_Internal_Sym *           sym;
      asection *                   sec;
      struct elf_link_hash_entry * h;
      bfd_vma                      relocation;
      bfd_reloc_status_type        r;
      arelent                      bfd_reloc;
      char                         sym_type;
      bfd_boolean                  unresolved_reloc = FALSE;
      char *error_message = NULL;
 
      r_symndx = ELF32_R_SYM (rel->r_info);
      r_type   = ELF32_R_TYPE (rel->r_info);
      r_type   = arm_real_reloc_type (globals, r_type);
 
      if (   r_type == R_ARM_GNU_VTENTRY
          || r_type == R_ARM_GNU_VTINHERIT)
        continue;
 
      bfd_reloc.howto = elf32_arm_howto_from_type (r_type);
      howto = bfd_reloc.howto;
 
      h = NULL;
      sym = NULL;
      sec = NULL;
 
      if (r_symndx < symtab_hdr->sh_info)
	{
	  sym = local_syms + r_symndx;
	  sym_type = ELF32_ST_TYPE (sym->st_info);
	  sec = local_sections[r_symndx];
 
	  /* An object file might have a reference to a local
	     undefined symbol.  This is a daft object file, but we
	     should at least do something about it.  V4BX & NONE
	     relocations do not use the symbol and are explicitly
	     allowed to use the undefined symbol, so allow those.
	     Likewise for relocations against STN_UNDEF.  */
	  if (r_type != R_ARM_V4BX
	      && r_type != R_ARM_NONE
	      && r_symndx != STN_UNDEF
	      && bfd_is_und_section (sec)
	      && ELF_ST_BIND (sym->st_info) != STB_WEAK)
	    {
	      if (!info->callbacks->undefined_symbol
		  (info, bfd_elf_string_from_elf_section
		   (input_bfd, symtab_hdr->sh_link, sym->st_name),
		   input_bfd, input_section,
		   rel->r_offset, TRUE))
		return FALSE;
	    }
 
	  if (globals->use_rel)
	    {
	      relocation = (sec->output_section->vma
			    + sec->output_offset
			    + sym->st_value);
	      if (!info->relocatable
		  && (sec->flags & SEC_MERGE)
		  && ELF_ST_TYPE (sym->st_info) == STT_SECTION)
		{
		  asection *msec;
		  bfd_vma addend, value;
 
		  switch (r_type)
		    {
		    case R_ARM_MOVW_ABS_NC:
		    case R_ARM_MOVT_ABS:
		      value = bfd_get_32 (input_bfd, contents + rel->r_offset);
		      addend = ((value & 0xf0000) >> 4) | (value & 0xfff);
		      addend = (addend ^ 0x8000) - 0x8000;
		      break;
 
		    case R_ARM_THM_MOVW_ABS_NC:
		    case R_ARM_THM_MOVT_ABS:
		      value = bfd_get_16 (input_bfd, contents + rel->r_offset)
			      << 16;
		      value |= bfd_get_16 (input_bfd,
					   contents + rel->r_offset + 2);
		      addend = ((value & 0xf7000) >> 4) | (value & 0xff)
			       | ((value & 0x04000000) >> 15);
		      addend = (addend ^ 0x8000) - 0x8000;
		      break;
 
		    default:
		      if (howto->rightshift
			  || (howto->src_mask & (howto->src_mask + 1)))
			{
			  (*_bfd_error_handler)
			    (_("%B(%A+0x%lx): %s relocation against SEC_MERGE section"),
			     input_bfd, input_section,
			     (long) rel->r_offset, howto->name);
			  return FALSE;
			}
 
		      value = bfd_get_32 (input_bfd, contents + rel->r_offset);
 
		      /* Get the (signed) value from the instruction.  */
		      addend = value & howto->src_mask;
		      if (addend & ((howto->src_mask + 1) >> 1))
			{
			  bfd_signed_vma mask;
 
			  mask = -1;
			  mask &= ~ howto->src_mask;
			  addend |= mask;
			}
		      break;
		    }
 
		  msec = sec;
		  addend =
		    _bfd_elf_rel_local_sym (output_bfd, sym, &msec, addend)
		    - relocation;
		  addend += msec->output_section->vma + msec->output_offset;
 
		  /* Cases here must match those in the preceding
		     switch statement.  */
		  switch (r_type)
		    {
		    case R_ARM_MOVW_ABS_NC:
		    case R_ARM_MOVT_ABS:
		      value = (value & 0xfff0f000) | ((addend & 0xf000) << 4)
			      | (addend & 0xfff);
		      bfd_put_32 (input_bfd, value, contents + rel->r_offset);
		      break;
 
		    case R_ARM_THM_MOVW_ABS_NC:
		    case R_ARM_THM_MOVT_ABS:
		      value = (value & 0xfbf08f00) | ((addend & 0xf700) << 4)
			      | (addend & 0xff) | ((addend & 0x0800) << 15);
		      bfd_put_16 (input_bfd, value >> 16,
				  contents + rel->r_offset);
		      bfd_put_16 (input_bfd, value,
				  contents + rel->r_offset + 2);
		      break;
 
		    default:
		      value = (value & ~ howto->dst_mask)
			      | (addend & howto->dst_mask);
		      bfd_put_32 (input_bfd, value, contents + rel->r_offset);
		      break;
		    }
		}
	    }
	  else
	    relocation = _bfd_elf_rela_local_sym (output_bfd, sym, &sec, rel);
	}
      else
	{
	  bfd_boolean warned;
 
	  RELOC_FOR_GLOBAL_SYMBOL (info, input_bfd, input_section, rel,
				   r_symndx, symtab_hdr, sym_hashes,
				   h, sec, relocation,
				   unresolved_reloc, warned);
 
	  sym_type = h->type;
	}
 
      if (sec != NULL && elf_discarded_section (sec))
	RELOC_AGAINST_DISCARDED_SECTION (info, input_bfd, input_section,
					 rel, relend, howto, contents);
 
      if (info->relocatable)
	{
	  /* This is a relocatable link.  We don't have to change
	     anything, unless the reloc is against a section symbol,
	     in which case we have to adjust according to where the
	     section symbol winds up in the output section.  */
	  if (sym != NULL && ELF_ST_TYPE (sym->st_info) == STT_SECTION)
	    {
	      if (globals->use_rel)
		arm_add_to_rel (input_bfd, contents + rel->r_offset,
				howto, (bfd_signed_vma) sec->output_offset);
	      else
		rel->r_addend += sec->output_offset;
	    }
	  continue;
	}
 
      if (h != NULL)
	name = h->root.root.string;
      else
	{
	  name = (bfd_elf_string_from_elf_section
		  (input_bfd, symtab_hdr->sh_link, sym->st_name));
	  if (name == NULL || *name == '\0')
	    name = bfd_section_name (input_bfd, sec);
	}
 
      if (r_symndx != STN_UNDEF
	  && r_type != R_ARM_NONE
	  && (h == NULL
	      || h->root.type == bfd_link_hash_defined
	      || h->root.type == bfd_link_hash_defweak)
	  && IS_ARM_TLS_RELOC (r_type) != (sym_type == STT_TLS))
	{
	  (*_bfd_error_handler)
	    ((sym_type == STT_TLS
	      ? _("%B(%A+0x%lx): %s used with TLS symbol %s")
	      : _("%B(%A+0x%lx): %s used with non-TLS symbol %s")),
	     input_bfd,
	     input_section,
	     (long) rel->r_offset,
	     howto->name,
	     name);
	}
 
      /* We call elf32_arm_final_link_relocate unless we're completely
         done, i.e., the relaxation produced the final output we want,
         and we won't let anybody mess with it. Also, we have to do
         addend adjustments in case of a R_ARM_TLS_GOTDESC relocation
         both in relaxed and non-relaxed cases */
     if ((elf32_arm_tls_transition (info, r_type, h) != (unsigned)r_type)
	 || (IS_ARM_TLS_GNU_RELOC (r_type)
	     && !((h ? elf32_arm_hash_entry (h)->tls_type : 
		   elf32_arm_local_got_tls_type (input_bfd)[r_symndx])
		  & GOT_TLS_GDESC)))
       {
	 r = elf32_arm_tls_relax (globals, input_bfd, input_section,
				  contents, rel, h == NULL);
	 /* This may have been marked unresolved because it came from
	    a shared library.  But we've just dealt with that.  */
	 unresolved_reloc = 0;
       }
     else
       r = bfd_reloc_continue;
 
     if (r == bfd_reloc_continue)
       r = elf32_arm_final_link_relocate (howto, input_bfd, output_bfd,
					  input_section, contents, rel,
					  relocation, info, sec, name, sym_type,
					  (h ? h->target_internal
					   : ARM_SYM_BRANCH_TYPE (sym)), h,
					  &unresolved_reloc, &error_message);
 
      /* Dynamic relocs are not propagated for SEC_DEBUGGING sections
	 because such sections are not SEC_ALLOC and thus ld.so will
	 not process them.  */
      if (unresolved_reloc
          && !((input_section->flags & SEC_DEBUGGING) != 0
               && h->def_dynamic))
	{
	  (*_bfd_error_handler)
	    (_("%B(%A+0x%lx): unresolvable %s relocation against symbol `%s'"),
	     input_bfd,
	     input_section,
	     (long) rel->r_offset,
	     howto->name,
	     h->root.root.string);
	  return FALSE;
	}
 
      if (r != bfd_reloc_ok)
	{
	  switch (r)
	    {
	    case bfd_reloc_overflow:
	      /* If the overflowing reloc was to an undefined symbol,
		 we have already printed one error message and there
		 is no point complaining again.  */
	      if ((! h ||
		   h->root.type != bfd_link_hash_undefined)
		  && (!((*info->callbacks->reloc_overflow)
			(info, (h ? &h->root : NULL), name, howto->name,
			 (bfd_vma) 0, input_bfd, input_section,
			 rel->r_offset))))
		  return FALSE;
	      break;
 
	    case bfd_reloc_undefined:
	      if (!((*info->callbacks->undefined_symbol)
		    (info, name, input_bfd, input_section,
		     rel->r_offset, TRUE)))
		return FALSE;
	      break;
 
	    case bfd_reloc_outofrange:
	      error_message = _("out of range");
	      goto common_error;
 
	    case bfd_reloc_notsupported:
	      error_message = _("unsupported relocation");
	      goto common_error;
 
	    case bfd_reloc_dangerous:
	      /* error_message should already be set.  */
	      goto common_error;
 
	    default:
	      error_message = _("unknown error");
	      /* Fall through.  */
 
	    common_error:
	      BFD_ASSERT (error_message != NULL);
	      if (!((*info->callbacks->reloc_dangerous)
		    (info, error_message, input_bfd, input_section,
		     rel->r_offset)))
		return FALSE;
	      break;
	    }
	}
    }
 
  return TRUE;
}
 
/* Add a new unwind edit to the list described by HEAD, TAIL.  If TINDEX is zero,
   adds the edit to the start of the list.  (The list must be built in order of
   ascending TINDEX: the function's callers are primarily responsible for
   maintaining that condition).  */
 
static void
add_unwind_table_edit (arm_unwind_table_edit **head,
		       arm_unwind_table_edit **tail,
		       arm_unwind_edit_type type,
		       asection *linked_section,
		       unsigned int tindex)
{
  arm_unwind_table_edit *new_edit = (arm_unwind_table_edit *)
      xmalloc (sizeof (arm_unwind_table_edit));
 
  new_edit->type = type;
  new_edit->linked_section = linked_section;
  new_edit->index = tindex;
 
  if (tindex > 0)
    {
      new_edit->next = NULL;
 
      if (*tail)
	(*tail)->next = new_edit;
 
      (*tail) = new_edit;
 
      if (!*head)
	(*head) = new_edit;
    }
  else
    {
      new_edit->next = *head;
 
      if (!*tail)
	*tail = new_edit;
 
      *head = new_edit;
    }
}
 
static _arm_elf_section_data *get_arm_elf_section_data (asection *);
 
/* Increase the size of EXIDX_SEC by ADJUST bytes.  ADJUST mau be negative.  */
static void
adjust_exidx_size(asection *exidx_sec, int adjust)
{
  asection *out_sec;
 
  if (!exidx_sec->rawsize)
    exidx_sec->rawsize = exidx_sec->size;
 
  bfd_set_section_size (exidx_sec->owner, exidx_sec, exidx_sec->size + adjust);
  out_sec = exidx_sec->output_section;
  /* Adjust size of output section.  */
  bfd_set_section_size (out_sec->owner, out_sec, out_sec->size +adjust);
}
 
/* Insert an EXIDX_CANTUNWIND marker at the end of a section.  */
static void
insert_cantunwind_after(asection *text_sec, asection *exidx_sec)
{
  struct _arm_elf_section_data *exidx_arm_data;
 
  exidx_arm_data = get_arm_elf_section_data (exidx_sec);
  add_unwind_table_edit (
    &exidx_arm_data->u.exidx.unwind_edit_list,
    &exidx_arm_data->u.exidx.unwind_edit_tail,
    INSERT_EXIDX_CANTUNWIND_AT_END, text_sec, UINT_MAX);
 
  adjust_exidx_size(exidx_sec, 8);
}
 
/* Scan .ARM.exidx tables, and create a list describing edits which should be
   made to those tables, such that:
 
     1. Regions without unwind data are marked with EXIDX_CANTUNWIND entries.
     2. Duplicate entries are merged together (EXIDX_CANTUNWIND, or unwind
        codes which have been inlined into the index).
 
   If MERGE_EXIDX_ENTRIES is false, duplicate entries are not merged.
 
   The edits are applied when the tables are written
   (in elf32_arm_write_section).
*/
 
bfd_boolean
elf32_arm_fix_exidx_coverage (asection **text_section_order,
			      unsigned int num_text_sections,
			      struct bfd_link_info *info,
			      bfd_boolean merge_exidx_entries)
{
  bfd *inp;
  unsigned int last_second_word = 0, i;
  asection *last_exidx_sec = NULL;
  asection *last_text_sec = NULL;
  int last_unwind_type = -1;
 
  /* Walk over all EXIDX sections, and create backlinks from the corrsponding
     text sections.  */
  for (inp = info->input_bfds; inp != NULL; inp = inp->link_next)
    {
      asection *sec;
 
      for (sec = inp->sections; sec != NULL; sec = sec->next)
        {
	  struct bfd_elf_section_data *elf_sec = elf_section_data (sec);
	  Elf_Internal_Shdr *hdr = &elf_sec->this_hdr;
 
	  if (!hdr || hdr->sh_type != SHT_ARM_EXIDX)
	    continue;
 
	  if (elf_sec->linked_to)
	    {
	      Elf_Internal_Shdr *linked_hdr
	        = &elf_section_data (elf_sec->linked_to)->this_hdr;
	      struct _arm_elf_section_data *linked_sec_arm_data
	        = get_arm_elf_section_data (linked_hdr->bfd_section);
 
	      if (linked_sec_arm_data == NULL)
	        continue;
 
	      /* Link this .ARM.exidx section back from the text section it
	         describes.  */
	      linked_sec_arm_data->u.text.arm_exidx_sec = sec;
	    }
	}
    }
 
  /* Walk all text sections in order of increasing VMA.  Eilminate duplicate
     index table entries (EXIDX_CANTUNWIND and inlined unwind opcodes),
     and add EXIDX_CANTUNWIND entries for sections with no unwind table data.  */
 
  for (i = 0; i < num_text_sections; i++)
    {
      asection *sec = text_section_order[i];
      asection *exidx_sec;
      struct _arm_elf_section_data *arm_data = get_arm_elf_section_data (sec);
      struct _arm_elf_section_data *exidx_arm_data;
      bfd_byte *contents = NULL;
      int deleted_exidx_bytes = 0;
      bfd_vma j;
      arm_unwind_table_edit *unwind_edit_head = NULL;
      arm_unwind_table_edit *unwind_edit_tail = NULL;
      Elf_Internal_Shdr *hdr;
      bfd *ibfd;
 
      if (arm_data == NULL)
        continue;
 
      exidx_sec = arm_data->u.text.arm_exidx_sec;
      if (exidx_sec == NULL)
	{
	  /* Section has no unwind data.  */
	  if (last_unwind_type == 0 || !last_exidx_sec)
	    continue;
 
	  /* Ignore zero sized sections.  */
	  if (sec->size == 0)
	    continue;
 
	  insert_cantunwind_after(last_text_sec, last_exidx_sec);
	  last_unwind_type = 0;
	  continue;
	}
 
      /* Skip /DISCARD/ sections.  */
      if (bfd_is_abs_section (exidx_sec->output_section))
	continue;
 
      hdr = &elf_section_data (exidx_sec)->this_hdr;
      if (hdr->sh_type != SHT_ARM_EXIDX)
        continue;
 
      exidx_arm_data = get_arm_elf_section_data (exidx_sec);
      if (exidx_arm_data == NULL)
        continue;
 
      ibfd = exidx_sec->owner;
 
      if (hdr->contents != NULL)
	contents = hdr->contents;
      else if (! bfd_malloc_and_get_section (ibfd, exidx_sec, &contents))
	/* An error?  */
	continue;
 
      for (j = 0; j < hdr->sh_size; j += 8)
	{
	  unsigned int second_word = bfd_get_32 (ibfd, contents + j + 4);
	  int unwind_type;
	  int elide = 0;
 
	  /* An EXIDX_CANTUNWIND entry.  */
	  if (second_word == 1)
	    {
	      if (last_unwind_type == 0)
		elide = 1;
	      unwind_type = 0;
	    }
	  /* Inlined unwinding data.  Merge if equal to previous.  */
	  else if ((second_word & 0x80000000) != 0)
	    {
	      if (merge_exidx_entries
		   && last_second_word == second_word && last_unwind_type == 1)
		elide = 1;
	      unwind_type = 1;
	      last_second_word = second_word;
	    }
	  /* Normal table entry.  In theory we could merge these too,
	     but duplicate entries are likely to be much less common.  */
	  else
	    unwind_type = 2;
 
	  if (elide)
	    {
	      add_unwind_table_edit (&unwind_edit_head, &unwind_edit_tail,
				     DELETE_EXIDX_ENTRY, NULL, j / 8);
 
	      deleted_exidx_bytes += 8;
	    }
 
	  last_unwind_type = unwind_type;
	}
 
      /* Free contents if we allocated it ourselves.  */
      if (contents != hdr->contents)
        free (contents);
 
      /* Record edits to be applied later (in elf32_arm_write_section).  */
      exidx_arm_data->u.exidx.unwind_edit_list = unwind_edit_head;
      exidx_arm_data->u.exidx.unwind_edit_tail = unwind_edit_tail;
 
      if (deleted_exidx_bytes > 0)
	adjust_exidx_size(exidx_sec, -deleted_exidx_bytes);
 
      last_exidx_sec = exidx_sec;
      last_text_sec = sec;
    }
 
  /* Add terminating CANTUNWIND entry.  */
  if (last_exidx_sec && last_unwind_type != 0)
    insert_cantunwind_after(last_text_sec, last_exidx_sec);
 
  return TRUE;
}
 
static bfd_boolean
elf32_arm_output_glue_section (struct bfd_link_info *info, bfd *obfd,
			       bfd *ibfd, const char *name)
{
  asection *sec, *osec;
 
  sec = bfd_get_section_by_name (ibfd, name);
  if (sec == NULL || (sec->flags & SEC_EXCLUDE) != 0)
    return TRUE;
 
  osec = sec->output_section;
  if (elf32_arm_write_section (obfd, info, sec, sec->contents))
    return TRUE;
 
  if (! bfd_set_section_contents (obfd, osec, sec->contents,
				  sec->output_offset, sec->size))
    return FALSE;
 
  return TRUE;
}
 
static bfd_boolean
elf32_arm_final_link (bfd *abfd, struct bfd_link_info *info)
{
  struct elf32_arm_link_hash_table *globals = elf32_arm_hash_table (info);
  asection *sec, *osec;
 
  if (globals == NULL)
    return FALSE;
 
  /* Invoke the regular ELF backend linker to do all the work.  */
  if (!bfd_elf_final_link (abfd, info))
    return FALSE;
 
  /* Process stub sections (eg BE8 encoding, ...).  */
  struct elf32_arm_link_hash_table *htab = elf32_arm_hash_table (info);
  int i;
  for (i=0; i<htab->top_id; i++)
    {
      sec = htab->stub_group[i].stub_sec;
      /* Only process it once, in its link_sec slot.  */
      if (sec && i == htab->stub_group[i].link_sec->id)
	{
	  osec = sec->output_section;
	  elf32_arm_write_section (abfd, info, sec, sec->contents);
	  if (! bfd_set_section_contents (abfd, osec, sec->contents,
					  sec->output_offset, sec->size))
	    return FALSE;
	}
    }
 
  /* Write out any glue sections now that we have created all the
     stubs.  */
  if (globals->bfd_of_glue_owner != NULL)
    {
      if (! elf32_arm_output_glue_section (info, abfd,
					   globals->bfd_of_glue_owner,
					   ARM2THUMB_GLUE_SECTION_NAME))
	return FALSE;
 
      if (! elf32_arm_output_glue_section (info, abfd,
					   globals->bfd_of_glue_owner,
					   THUMB2ARM_GLUE_SECTION_NAME))
	return FALSE;
 
      if (! elf32_arm_output_glue_section (info, abfd,
					   globals->bfd_of_glue_owner,
					   VFP11_ERRATUM_VENEER_SECTION_NAME))
	return FALSE;
 
      if (! elf32_arm_output_glue_section (info, abfd,
					   globals->bfd_of_glue_owner,
					   ARM_BX_GLUE_SECTION_NAME))
	return FALSE;
    }
 
  return TRUE;
}
 
/* Set the right machine number.  */
 
static bfd_boolean
elf32_arm_object_p (bfd *abfd)
{
  unsigned int mach;
 
  mach = bfd_arm_get_mach_from_notes (abfd, ARM_NOTE_SECTION);
 
  if (mach != bfd_mach_arm_unknown)
    bfd_default_set_arch_mach (abfd, bfd_arch_arm, mach);
 
  else if (elf_elfheader (abfd)->e_flags & EF_ARM_MAVERICK_FLOAT)
    bfd_default_set_arch_mach (abfd, bfd_arch_arm, bfd_mach_arm_ep9312);
 
  else
    bfd_default_set_arch_mach (abfd, bfd_arch_arm, mach);
 
  return TRUE;
}
 
/* Function to keep ARM specific flags in the ELF header.  */
 
static bfd_boolean
elf32_arm_set_private_flags (bfd *abfd, flagword flags)
{
  if (elf_flags_init (abfd)
      && elf_elfheader (abfd)->e_flags != flags)
    {
      if (EF_ARM_EABI_VERSION (flags) == EF_ARM_EABI_UNKNOWN)
	{
	  if (flags & EF_ARM_INTERWORK)
	    (*_bfd_error_handler)
	      (_("Warning: Not setting interworking flag of %B since it has already been specified as non-interworking"),
	       abfd);
	  else
	    _bfd_error_handler
	      (_("Warning: Clearing the interworking flag of %B due to outside request"),
	       abfd);
	}
    }
  else
    {
      elf_elfheader (abfd)->e_flags = flags;
      elf_flags_init (abfd) = TRUE;
    }
 
  return TRUE;
}
 
/* Copy backend specific data from one object module to another.  */
 
static bfd_boolean
elf32_arm_copy_private_bfd_data (bfd *ibfd, bfd *obfd)
{
  flagword in_flags;
  flagword out_flags;
 
  if (! is_arm_elf (ibfd) || ! is_arm_elf (obfd))
    return TRUE;
 
  in_flags  = elf_elfheader (ibfd)->e_flags;
  out_flags = elf_elfheader (obfd)->e_flags;
 
  if (elf_flags_init (obfd)
      && EF_ARM_EABI_VERSION (out_flags) == EF_ARM_EABI_UNKNOWN
      && in_flags != out_flags)
    {
      /* Cannot mix APCS26 and APCS32 code.  */
      if ((in_flags & EF_ARM_APCS_26) != (out_flags & EF_ARM_APCS_26))
	return FALSE;
 
      /* Cannot mix float APCS and non-float APCS code.  */
      if ((in_flags & EF_ARM_APCS_FLOAT) != (out_flags & EF_ARM_APCS_FLOAT))
	return FALSE;
 
      /* If the src and dest have different interworking flags
         then turn off the interworking bit.  */
      if ((in_flags & EF_ARM_INTERWORK) != (out_flags & EF_ARM_INTERWORK))
	{
	  if (out_flags & EF_ARM_INTERWORK)
	    _bfd_error_handler
	      (_("Warning: Clearing the interworking flag of %B because non-interworking code in %B has been linked with it"),
	       obfd, ibfd);
 
	  in_flags &= ~EF_ARM_INTERWORK;
	}
 
      /* Likewise for PIC, though don't warn for this case.  */
      if ((in_flags & EF_ARM_PIC) != (out_flags & EF_ARM_PIC))
	in_flags &= ~EF_ARM_PIC;
    }
 
  elf_elfheader (obfd)->e_flags = in_flags;
  elf_flags_init (obfd) = TRUE;
 
  /* Also copy the EI_OSABI field.  */
  elf_elfheader (obfd)->e_ident[EI_OSABI] =
    elf_elfheader (ibfd)->e_ident[EI_OSABI];
 
  /* Copy object attributes.  */
  _bfd_elf_copy_obj_attributes (ibfd, obfd);
 
  return TRUE;
}
 
/* Values for Tag_ABI_PCS_R9_use.  */
enum
{
  AEABI_R9_V6,
  AEABI_R9_SB,
  AEABI_R9_TLS,
  AEABI_R9_unused
};
 
/* Values for Tag_ABI_PCS_RW_data.  */
enum
{
  AEABI_PCS_RW_data_absolute,
  AEABI_PCS_RW_data_PCrel,
  AEABI_PCS_RW_data_SBrel,
  AEABI_PCS_RW_data_unused
};
 
/* Values for Tag_ABI_enum_size.  */
enum
{
  AEABI_enum_unused,
  AEABI_enum_short,
  AEABI_enum_wide,
  AEABI_enum_forced_wide
};
 
/* Determine whether an object attribute tag takes an integer, a
   string or both.  */
 
static int
elf32_arm_obj_attrs_arg_type (int tag)
{
  if (tag == Tag_compatibility)
    return ATTR_TYPE_FLAG_INT_VAL | ATTR_TYPE_FLAG_STR_VAL;
  else if (tag == Tag_nodefaults)
    return ATTR_TYPE_FLAG_INT_VAL | ATTR_TYPE_FLAG_NO_DEFAULT;
  else if (tag == Tag_CPU_raw_name || tag == Tag_CPU_name)
    return ATTR_TYPE_FLAG_STR_VAL;
  else if (tag < 32)
    return ATTR_TYPE_FLAG_INT_VAL;
  else
    return (tag & 1) != 0 ? ATTR_TYPE_FLAG_STR_VAL : ATTR_TYPE_FLAG_INT_VAL;
}
 
/* The ABI defines that Tag_conformance should be emitted first, and that
   Tag_nodefaults should be second (if either is defined).  This sets those
   two positions, and bumps up the position of all the remaining tags to
   compensate.  */
static int
elf32_arm_obj_attrs_order (int num)
{
  if (num == LEAST_KNOWN_OBJ_ATTRIBUTE)
    return Tag_conformance;
  if (num == LEAST_KNOWN_OBJ_ATTRIBUTE + 1)
    return Tag_nodefaults;
  if ((num - 2) < Tag_nodefaults)
    return num - 2;
  if ((num - 1) < Tag_conformance)
    return num - 1;
  return num;
}
 
/* Attribute numbers >=64 (mod 128) can be safely ignored.  */
static bfd_boolean
elf32_arm_obj_attrs_handle_unknown (bfd *abfd, int tag)
{
  if ((tag & 127) < 64)
    {
      _bfd_error_handler
	(_("%B: Unknown mandatory EABI object attribute %d"),
	 abfd, tag);
      bfd_set_error (bfd_error_bad_value);
      return FALSE;
    }
  else
    {
      _bfd_error_handler
	(_("Warning: %B: Unknown EABI object attribute %d"),
	 abfd, tag);
      return TRUE;
    }
}
 
/* Read the architecture from the Tag_also_compatible_with attribute, if any.
   Returns -1 if no architecture could be read.  */
 
static int
get_secondary_compatible_arch (bfd *abfd)
{
  obj_attribute *attr =
    &elf_known_obj_attributes_proc (abfd)[Tag_also_compatible_with];
 
  /* Note: the tag and its argument below are uleb128 values, though
     currently-defined values fit in one byte for each.  */
  if (attr->s
      && attr->s[0] == Tag_CPU_arch
      && (attr->s[1] & 128) != 128
      && attr->s[2] == 0)
   return attr->s[1];
 
  /* This tag is "safely ignorable", so don't complain if it looks funny.  */
  return -1;
}
 
/* Set, or unset, the architecture of the Tag_also_compatible_with attribute.
   The tag is removed if ARCH is -1.  */
 
static void
set_secondary_compatible_arch (bfd *abfd, int arch)
{
  obj_attribute *attr =
    &elf_known_obj_attributes_proc (abfd)[Tag_also_compatible_with];
 
  if (arch == -1)
    {
      attr->s = NULL;
      return;
    }
 
  /* Note: the tag and its argument below are uleb128 values, though
     currently-defined values fit in one byte for each.  */
  if (!attr->s)
    attr->s = (char *) bfd_alloc (abfd, 3);
  attr->s[0] = Tag_CPU_arch;
  attr->s[1] = arch;
  attr->s[2] = '\0';
}
 
/* Combine two values for Tag_CPU_arch, taking secondary compatibility tags
   into account.  */
 
static int
tag_cpu_arch_combine (bfd *ibfd, int oldtag, int *secondary_compat_out,
		      int newtag, int secondary_compat)
{
#define T(X) TAG_CPU_ARCH_##X
  int tagl, tagh, result;
  const int v6t2[] =
    {
      T(V6T2),   /* PRE_V4.  */
      T(V6T2),   /* V4.  */
      T(V6T2),   /* V4T.  */
      T(V6T2),   /* V5T.  */
      T(V6T2),   /* V5TE.  */
      T(V6T2),   /* V5TEJ.  */
      T(V6T2),   /* V6.  */
      T(V7),     /* V6KZ.  */
      T(V6T2)    /* V6T2.  */
    };
  const int v6k[] =
    {
      T(V6K),    /* PRE_V4.  */
      T(V6K),    /* V4.  */
      T(V6K),    /* V4T.  */
      T(V6K),    /* V5T.  */
      T(V6K),    /* V5TE.  */
      T(V6K),    /* V5TEJ.  */
      T(V6K),    /* V6.  */
      T(V6KZ),   /* V6KZ.  */
      T(V7),     /* V6T2.  */
      T(V6K)     /* V6K.  */
    };
  const int v7[] =
    {
      T(V7),     /* PRE_V4.  */
      T(V7),     /* V4.  */
      T(V7),     /* V4T.  */
      T(V7),     /* V5T.  */
      T(V7),     /* V5TE.  */
      T(V7),     /* V5TEJ.  */
      T(V7),     /* V6.  */
      T(V7),     /* V6KZ.  */
      T(V7),     /* V6T2.  */
      T(V7),     /* V6K.  */
      T(V7)      /* V7.  */
    };
  const int v6_m[] =
    {
      -1,        /* PRE_V4.  */
      -1,        /* V4.  */
      T(V6K),    /* V4T.  */
      T(V6K),    /* V5T.  */
      T(V6K),    /* V5TE.  */
      T(V6K),    /* V5TEJ.  */
      T(V6K),    /* V6.  */
      T(V6KZ),   /* V6KZ.  */
      T(V7),     /* V6T2.  */
      T(V6K),    /* V6K.  */
      T(V7),     /* V7.  */
      T(V6_M)    /* V6_M.  */
    };
  const int v6s_m[] =
    {
      -1,        /* PRE_V4.  */
      -1,        /* V4.  */
      T(V6K),    /* V4T.  */
      T(V6K),    /* V5T.  */
      T(V6K),    /* V5TE.  */
      T(V6K),    /* V5TEJ.  */
      T(V6K),    /* V6.  */
      T(V6KZ),   /* V6KZ.  */
      T(V7),     /* V6T2.  */
      T(V6K),    /* V6K.  */
      T(V7),     /* V7.  */
      T(V6S_M),  /* V6_M.  */
      T(V6S_M)   /* V6S_M.  */
    };
  const int v7e_m[] =
    {
      -1,        /* PRE_V4.  */
      -1,        /* V4.  */
      T(V7E_M),  /* V4T.  */
      T(V7E_M),  /* V5T.  */
      T(V7E_M),  /* V5TE.  */
      T(V7E_M),  /* V5TEJ.  */
      T(V7E_M),  /* V6.  */
      T(V7E_M),  /* V6KZ.  */
      T(V7E_M),  /* V6T2.  */
      T(V7E_M),  /* V6K.  */
      T(V7E_M),  /* V7.  */
      T(V7E_M),  /* V6_M.  */
      T(V7E_M),  /* V6S_M.  */
      T(V7E_M)   /* V7E_M.  */
    };
  const int v4t_plus_v6_m[] =
    {
      -1,		/* PRE_V4.  */
      -1,		/* V4.  */
      T(V4T),		/* V4T.  */
      T(V5T),		/* V5T.  */
      T(V5TE),		/* V5TE.  */
      T(V5TEJ),		/* V5TEJ.  */
      T(V6),		/* V6.  */
      T(V6KZ),		/* V6KZ.  */
      T(V6T2),		/* V6T2.  */
      T(V6K),		/* V6K.  */
      T(V7),		/* V7.  */
      T(V6_M),		/* V6_M.  */
      T(V6S_M),		/* V6S_M.  */
      T(V7E_M),		/* V7E_M.  */
      T(V4T_PLUS_V6_M)	/* V4T plus V6_M.  */
    };
  const int *comb[] =
    {
      v6t2,
      v6k,
      v7,
      v6_m,
      v6s_m,
      v7e_m,
      /* Pseudo-architecture.  */
      v4t_plus_v6_m
    };
 
  /* Check we've not got a higher architecture than we know about.  */
 
  if (oldtag > MAX_TAG_CPU_ARCH || newtag > MAX_TAG_CPU_ARCH)
    {
      _bfd_error_handler (_("error: %B: Unknown CPU architecture"), ibfd);
      return -1;
    }
 
  /* Override old tag if we have a Tag_also_compatible_with on the output.  */
 
  if ((oldtag == T(V6_M) && *secondary_compat_out == T(V4T))
      || (oldtag == T(V4T) && *secondary_compat_out == T(V6_M)))
    oldtag = T(V4T_PLUS_V6_M);
 
  /* And override the new tag if we have a Tag_also_compatible_with on the
     input.  */
 
  if ((newtag == T(V6_M) && secondary_compat == T(V4T))
      || (newtag == T(V4T) && secondary_compat == T(V6_M)))
    newtag = T(V4T_PLUS_V6_M);
 
  tagl = (oldtag < newtag) ? oldtag : newtag;
  result = tagh = (oldtag > newtag) ? oldtag : newtag;
 
  /* Architectures before V6KZ add features monotonically.  */
  if (tagh <= TAG_CPU_ARCH_V6KZ)
    return result;
 
  result = comb[tagh - T(V6T2)][tagl];
 
  /* Use Tag_CPU_arch == V4T and Tag_also_compatible_with (Tag_CPU_arch V6_M)
     as the canonical version.  */
  if (result == T(V4T_PLUS_V6_M))
    {
      result = T(V4T);
      *secondary_compat_out = T(V6_M);
    }
  else
    *secondary_compat_out = -1;
 
  if (result == -1)
    {
      _bfd_error_handler (_("error: %B: Conflicting CPU architectures %d/%d"),
			  ibfd, oldtag, newtag);
      return -1;
    }
 
  return result;
#undef T
}
 
/* Merge EABI object attributes from IBFD into OBFD.  Raise an error if there
   are conflicting attributes.  */
 
static bfd_boolean
elf32_arm_merge_eabi_attributes (bfd *ibfd, bfd *obfd)
{
  obj_attribute *in_attr;
  obj_attribute *out_attr;
  /* Some tags have 0 = don't care, 1 = strong requirement,
     2 = weak requirement.  */
  static const int order_021[3] = {0, 2, 1};
  int i;
  bfd_boolean result = TRUE;
 
  /* Skip the linker stubs file.  This preserves previous behavior
     of accepting unknown attributes in the first input file - but
     is that a bug?  */
  if (ibfd->flags & BFD_LINKER_CREATED)
    return TRUE;
 
  if (!elf_known_obj_attributes_proc (obfd)[0].i)
    {
      /* This is the first object.  Copy the attributes.  */
      _bfd_elf_copy_obj_attributes (ibfd, obfd);
 
      out_attr = elf_known_obj_attributes_proc (obfd);
 
      /* Use the Tag_null value to indicate the attributes have been
	 initialized.  */
      out_attr[0].i = 1;
 
      /* We do not output objects with Tag_MPextension_use_legacy - we move
	 the attribute's value to Tag_MPextension_use.  */
      if (out_attr[Tag_MPextension_use_legacy].i != 0)
	{
	  if (out_attr[Tag_MPextension_use].i != 0
	      && out_attr[Tag_MPextension_use_legacy].i
	        != out_attr[Tag_MPextension_use].i)
	    {
	      _bfd_error_handler
		(_("Error: %B has both the current and legacy "
		   "Tag_MPextension_use attributes"), ibfd);
	      result = FALSE;
	    }
 
	  out_attr[Tag_MPextension_use] =
	    out_attr[Tag_MPextension_use_legacy];
	  out_attr[Tag_MPextension_use_legacy].type = 0;
	  out_attr[Tag_MPextension_use_legacy].i = 0;
	}
 
      return result;
    }
 
  in_attr = elf_known_obj_attributes_proc (ibfd);
  out_attr = elf_known_obj_attributes_proc (obfd);
  /* This needs to happen before Tag_ABI_FP_number_model is merged.  */
  if (in_attr[Tag_ABI_VFP_args].i != out_attr[Tag_ABI_VFP_args].i)
    {
      /* Ignore mismatches if the object doesn't use floating point.  */
      if (out_attr[Tag_ABI_FP_number_model].i == 0)
	out_attr[Tag_ABI_VFP_args].i = in_attr[Tag_ABI_VFP_args].i;
      else if (in_attr[Tag_ABI_FP_number_model].i != 0)
	{
	  _bfd_error_handler
	    (_("error: %B uses VFP register arguments, %B does not"),
	     in_attr[Tag_ABI_VFP_args].i ? ibfd : obfd,
	     in_attr[Tag_ABI_VFP_args].i ? obfd : ibfd);
	  result = FALSE;
	}
    }
 
  for (i = LEAST_KNOWN_OBJ_ATTRIBUTE; i < NUM_KNOWN_OBJ_ATTRIBUTES; i++)
    {
      /* Merge this attribute with existing attributes.  */
      switch (i)
	{
	case Tag_CPU_raw_name:
	case Tag_CPU_name:
	  /* These are merged after Tag_CPU_arch. */
	  break;
 
	case Tag_ABI_optimization_goals:
	case Tag_ABI_FP_optimization_goals:
	  /* Use the first value seen.  */
	  break;
 
	case Tag_CPU_arch:
	  {
	    int secondary_compat = -1, secondary_compat_out = -1;
	    unsigned int saved_out_attr = out_attr[i].i;
	    static const char *name_table[] = {
		/* These aren't real CPU names, but we can't guess
		   that from the architecture version alone.  */
		"Pre v4",
		"ARM v4",
		"ARM v4T",
		"ARM v5T",
		"ARM v5TE",
		"ARM v5TEJ",
		"ARM v6",
		"ARM v6KZ",
		"ARM v6T2",
		"ARM v6K",
		"ARM v7",
		"ARM v6-M",
		"ARM v6S-M"
	    };
 
	    /* Merge Tag_CPU_arch and Tag_also_compatible_with.  */
	    secondary_compat = get_secondary_compatible_arch (ibfd);
	    secondary_compat_out = get_secondary_compatible_arch (obfd);
	    out_attr[i].i = tag_cpu_arch_combine (ibfd, out_attr[i].i,
						  &secondary_compat_out,
						  in_attr[i].i,
						  secondary_compat);
	    set_secondary_compatible_arch (obfd, secondary_compat_out);
 
	    /* Merge Tag_CPU_name and Tag_CPU_raw_name.  */
	    if (out_attr[i].i == saved_out_attr)
	      ; /* Leave the names alone.  */
	    else if (out_attr[i].i == in_attr[i].i)
	      {
		/* The output architecture has been changed to match the
		   input architecture.  Use the input names.  */
		out_attr[Tag_CPU_name].s = in_attr[Tag_CPU_name].s
		  ? _bfd_elf_attr_strdup (obfd, in_attr[Tag_CPU_name].s)
		  : NULL;
		out_attr[Tag_CPU_raw_name].s = in_attr[Tag_CPU_raw_name].s
		  ? _bfd_elf_attr_strdup (obfd, in_attr[Tag_CPU_raw_name].s)
		  : NULL;
	      }
	    else
	      {
		out_attr[Tag_CPU_name].s = NULL;
		out_attr[Tag_CPU_raw_name].s = NULL;
	      }
 
	    /* If we still don't have a value for Tag_CPU_name,
	       make one up now.  Tag_CPU_raw_name remains blank.  */
	    if (out_attr[Tag_CPU_name].s == NULL
		&& out_attr[i].i < ARRAY_SIZE (name_table))
	      out_attr[Tag_CPU_name].s =
		_bfd_elf_attr_strdup (obfd, name_table[out_attr[i].i]);
	  }
	  break;
 
	case Tag_ARM_ISA_use:
	case Tag_THUMB_ISA_use:
	case Tag_WMMX_arch:
	case Tag_Advanced_SIMD_arch:
	  /* ??? Do Advanced_SIMD (NEON) and WMMX conflict?  */
	case Tag_ABI_FP_rounding:
	case Tag_ABI_FP_exceptions:
	case Tag_ABI_FP_user_exceptions:
	case Tag_ABI_FP_number_model:
	case Tag_FP_HP_extension:
	case Tag_CPU_unaligned_access:
	case Tag_T2EE_use:
	case Tag_MPextension_use:
	  /* Use the largest value specified.  */
	  if (in_attr[i].i > out_attr[i].i)
	    out_attr[i].i = in_attr[i].i;
	  break;
 
	case Tag_ABI_align_preserved:
	case Tag_ABI_PCS_RO_data:
	  /* Use the smallest value specified.  */
	  if (in_attr[i].i < out_attr[i].i)
	    out_attr[i].i = in_attr[i].i;
	  break;
 
	case Tag_ABI_align_needed:
	  if ((in_attr[i].i > 0 || out_attr[i].i > 0)
	      && (in_attr[Tag_ABI_align_preserved].i == 0
		  || out_attr[Tag_ABI_align_preserved].i == 0))
	    {
	      /* This error message should be enabled once all non-conformant
		 binaries in the toolchain have had the attributes set
		 properly.
	      _bfd_error_handler
		(_("error: %B: 8-byte data alignment conflicts with %B"),
		 obfd, ibfd);
	      result = FALSE; */
	    }
	  /* Fall through.  */
	case Tag_ABI_FP_denormal:
	case Tag_ABI_PCS_GOT_use:
	  /* Use the "greatest" from the sequence 0, 2, 1, or the largest
	     value if greater than 2 (for future-proofing).  */
	  if ((in_attr[i].i > 2 && in_attr[i].i > out_attr[i].i)
	      || (in_attr[i].i <= 2 && out_attr[i].i <= 2
		  && order_021[in_attr[i].i] > order_021[out_attr[i].i]))
	    out_attr[i].i = in_attr[i].i;
	  break;
 
	case Tag_Virtualization_use:
	  /* The virtualization tag effectively stores two bits of
	     information: the intended use of TrustZone (in bit 0), and the
	     intended use of Virtualization (in bit 1).  */
	  if (out_attr[i].i == 0)
	    out_attr[i].i = in_attr[i].i;
	  else if (in_attr[i].i != 0
		   && in_attr[i].i != out_attr[i].i)
	    {
	      if (in_attr[i].i <= 3 && out_attr[i].i <= 3)
		out_attr[i].i = 3;
	      else
		{
		  _bfd_error_handler
		    (_("error: %B: unable to merge virtualization attributes "
		       "with %B"),
		     obfd, ibfd);
		  result = FALSE;
		}
	    }
	  break;
 
	case Tag_CPU_arch_profile:
	  if (out_attr[i].i != in_attr[i].i)
	    {
	      /* 0 will merge with anything.
		 'A' and 'S' merge to 'A'.
		 'R' and 'S' merge to 'R'.
	         'M' and 'A|R|S' is an error.  */
	      if (out_attr[i].i == 0
		  || (out_attr[i].i == 'S'
		      && (in_attr[i].i == 'A' || in_attr[i].i == 'R')))
		out_attr[i].i = in_attr[i].i;
	      else if (in_attr[i].i == 0
		       || (in_attr[i].i == 'S'
			   && (out_attr[i].i == 'A' || out_attr[i].i == 'R')))
		; /* Do nothing. */
	      else
		{
		  _bfd_error_handler
		    (_("error: %B: Conflicting architecture profiles %c/%c"),
		     ibfd,
		     in_attr[i].i ? in_attr[i].i : '0',
		     out_attr[i].i ? out_attr[i].i : '0');
		  result = FALSE;
		}
	    }
	  break;
	case Tag_FP_arch:
	    {
	      /* Tag_ABI_HardFP_use is handled along with Tag_FP_arch since
		 the meaning of Tag_ABI_HardFP_use depends on Tag_FP_arch
		 when it's 0.  It might mean absence of FP hardware if
		 Tag_FP_arch is zero, otherwise it is effectively SP + DP.  */
 
	      static const struct
	      {
		  int ver;
		  int regs;
	      } vfp_versions[7] =
		{
		  {0, 0},
		  {1, 16},
		  {2, 16},
		  {3, 32},
		  {3, 16},
		  {4, 32},
		  {4, 16}
		};
	      int ver;
	      int regs;
	      int newval;
 
	      /* If the output has no requirement about FP hardware,
		 follow the requirement of the input.  */
	      if (out_attr[i].i == 0)
		{
		  BFD_ASSERT (out_attr[Tag_ABI_HardFP_use].i == 0);
		  out_attr[i].i = in_attr[i].i;
		  out_attr[Tag_ABI_HardFP_use].i
		    = in_attr[Tag_ABI_HardFP_use].i;
		  break;
		}
	      /* If the input has no requirement about FP hardware, do
		 nothing.  */
	      else if (in_attr[i].i == 0)
		{
		  BFD_ASSERT (in_attr[Tag_ABI_HardFP_use].i == 0);
		  break;
		}
 
	      /* Both the input and the output have nonzero Tag_FP_arch.
		 So Tag_ABI_HardFP_use is (SP & DP) when it's zero.  */
 
	      /* If both the input and the output have zero Tag_ABI_HardFP_use,
		 do nothing.  */
	      if (in_attr[Tag_ABI_HardFP_use].i == 0
		  && out_attr[Tag_ABI_HardFP_use].i == 0)
		;
	      /* If the input and the output have different Tag_ABI_HardFP_use,
		 the combination of them is 3 (SP & DP).  */
	      else if (in_attr[Tag_ABI_HardFP_use].i
		       != out_attr[Tag_ABI_HardFP_use].i)
		out_attr[Tag_ABI_HardFP_use].i = 3;
 
	      /* Now we can handle Tag_FP_arch.  */
 
	      /* Values greater than 6 aren't defined, so just pick the
	         biggest */
	      if (in_attr[i].i > 6 && in_attr[i].i > out_attr[i].i)
		{
		  out_attr[i] = in_attr[i];
		  break;
		}
	      /* The output uses the superset of input features
		 (ISA version) and registers.  */
	      ver = vfp_versions[in_attr[i].i].ver;
	      if (ver < vfp_versions[out_attr[i].i].ver)
		ver = vfp_versions[out_attr[i].i].ver;
	      regs = vfp_versions[in_attr[i].i].regs;
	      if (regs < vfp_versions[out_attr[i].i].regs)
		regs = vfp_versions[out_attr[i].i].regs;
	      /* This assumes all possible supersets are also a valid
	         options.  */
	      for (newval = 6; newval > 0; newval--)
		{
		  if (regs == vfp_versions[newval].regs
		      && ver == vfp_versions[newval].ver)
		    break;
		}
	      out_attr[i].i = newval;
	    }
	  break;
	case Tag_PCS_config:
	  if (out_attr[i].i == 0)
	    out_attr[i].i = in_attr[i].i;
	  else if (in_attr[i].i != 0 && out_attr[i].i != 0)
	    {
	      /* It's sometimes ok to mix different configs, so this is only
	         a warning.  */
	      _bfd_error_handler
		(_("Warning: %B: Conflicting platform configuration"), ibfd);
	    }
	  break;
	case Tag_ABI_PCS_R9_use:
	  if (in_attr[i].i != out_attr[i].i
	      && out_attr[i].i != AEABI_R9_unused
	      && in_attr[i].i != AEABI_R9_unused)
	    {
	      _bfd_error_handler
		(_("error: %B: Conflicting use of R9"), ibfd);
	      result = FALSE;
	    }
	  if (out_attr[i].i == AEABI_R9_unused)
	    out_attr[i].i = in_attr[i].i;
	  break;
	case Tag_ABI_PCS_RW_data:
	  if (in_attr[i].i == AEABI_PCS_RW_data_SBrel
	      && out_attr[Tag_ABI_PCS_R9_use].i != AEABI_R9_SB
	      && out_attr[Tag_ABI_PCS_R9_use].i != AEABI_R9_unused)
	    {
	      _bfd_error_handler
		(_("error: %B: SB relative addressing conflicts with use of R9"),
		 ibfd);
	      result = FALSE;
	    }
	  /* Use the smallest value specified.  */
	  if (in_attr[i].i < out_attr[i].i)
	    out_attr[i].i = in_attr[i].i;
	  break;
	case Tag_ABI_PCS_wchar_t:
	  if (out_attr[i].i && in_attr[i].i && out_attr[i].i != in_attr[i].i
	      && !elf_arm_tdata (obfd)->no_wchar_size_warning)
	    {
	      _bfd_error_handler
		(_("warning: %B uses %u-byte wchar_t yet the output is to use %u-byte wchar_t; use of wchar_t values across objects may fail"),
		 ibfd, in_attr[i].i, out_attr[i].i);
	    }
	  else if (in_attr[i].i && !out_attr[i].i)
	    out_attr[i].i = in_attr[i].i;
	  break;
	case Tag_ABI_enum_size:
	  if (in_attr[i].i != AEABI_enum_unused)
	    {
	      if (out_attr[i].i == AEABI_enum_unused
		  || out_attr[i].i == AEABI_enum_forced_wide)
		{
		  /* The existing object is compatible with anything.
		     Use whatever requirements the new object has.  */
		  out_attr[i].i = in_attr[i].i;
		}
	      else if (in_attr[i].i != AEABI_enum_forced_wide
		       && out_attr[i].i != in_attr[i].i
		       && !elf_arm_tdata (obfd)->no_enum_size_warning)
		{
		  static const char *aeabi_enum_names[] =
		    { "", "variable-size", "32-bit", "" };
		  const char *in_name =
		    in_attr[i].i < ARRAY_SIZE(aeabi_enum_names)
		    ? aeabi_enum_names[in_attr[i].i]
		    : "<unknown>";
		  const char *out_name =
		    out_attr[i].i < ARRAY_SIZE(aeabi_enum_names)
		    ? aeabi_enum_names[out_attr[i].i]
		    : "<unknown>";
		  _bfd_error_handler
		    (_("warning: %B uses %s enums yet the output is to use %s enums; use of enum values across objects may fail"),
		     ibfd, in_name, out_name);
		}
	    }
	  break;
	case Tag_ABI_VFP_args:
	  /* Aready done.  */
	  break;
	case Tag_ABI_WMMX_args:
	  if (in_attr[i].i != out_attr[i].i)
	    {
	      _bfd_error_handler
		(_("error: %B uses iWMMXt register arguments, %B does not"),
		 ibfd, obfd);
	      result = FALSE;
	    }
	  break;
	case Tag_compatibility:
	  /* Merged in target-independent code.  */
	  break;
	case Tag_ABI_HardFP_use:
	  /* This is handled along with Tag_FP_arch.  */
	  break;
	case Tag_ABI_FP_16bit_format:
	  if (in_attr[i].i != 0 && out_attr[i].i != 0)
	    {
	      if (in_attr[i].i != out_attr[i].i)
		{
		  _bfd_error_handler
		    (_("error: fp16 format mismatch between %B and %B"),
		     ibfd, obfd);
		  result = FALSE;
		}
	    }
	  if (in_attr[i].i != 0)
	    out_attr[i].i = in_attr[i].i;
	  break;
 
	case Tag_DIV_use:
	  /* This tag is set to zero if we can use UDIV and SDIV in Thumb
	     mode on a v7-M or v7-R CPU; to one if we can not use UDIV or
	     SDIV at all; and to two if we can use UDIV or SDIV on a v7-A
	     CPU.  We will merge as follows: If the input attribute's value
	     is one then the output attribute's value remains unchanged.  If
	     the input attribute's value is zero or two then if the output
	     attribute's value is one the output value is set to the input
	     value, otherwise the output value must be the same as the
	     inputs.  */ 
	  if (in_attr[i].i != 1 && out_attr[i].i != 1) 
	    { 
	      if (in_attr[i].i != out_attr[i].i)
		{
		  _bfd_error_handler
		    (_("DIV usage mismatch between %B and %B"),
		     ibfd, obfd); 
		  result = FALSE;
		}
	    } 
 
	  if (in_attr[i].i != 1)
	    out_attr[i].i = in_attr[i].i; 
 
	  break;
 
	case Tag_MPextension_use_legacy:
	  /* We don't output objects with Tag_MPextension_use_legacy - we
	     move the value to Tag_MPextension_use.  */
	  if (in_attr[i].i != 0 && in_attr[Tag_MPextension_use].i != 0)
	    {
	      if (in_attr[Tag_MPextension_use].i != in_attr[i].i)
		{
		  _bfd_error_handler
		    (_("%B has has both the current and legacy "
		       "Tag_MPextension_use attributes"), 
		     ibfd);
		  result = FALSE;
		}
	    }
 
	  if (in_attr[i].i > out_attr[Tag_MPextension_use].i)
	    out_attr[Tag_MPextension_use] = in_attr[i];
 
	  break;
 
	case Tag_nodefaults:
	  /* This tag is set if it exists, but the value is unused (and is
	     typically zero).  We don't actually need to do anything here -
	     the merge happens automatically when the type flags are merged
	     below.  */
	  break;
	case Tag_also_compatible_with:
	  /* Already done in Tag_CPU_arch.  */
	  break;
	case Tag_conformance:
	  /* Keep the attribute if it matches.  Throw it away otherwise.
	     No attribute means no claim to conform.  */
	  if (!in_attr[i].s || !out_attr[i].s
	      || strcmp (in_attr[i].s, out_attr[i].s) != 0)
	    out_attr[i].s = NULL;
	  break;
 
	default:
	  result
	    = result && _bfd_elf_merge_unknown_attribute_low (ibfd, obfd, i);
	}
 
      /* If out_attr was copied from in_attr then it won't have a type yet.  */
      if (in_attr[i].type && !out_attr[i].type)
	out_attr[i].type = in_attr[i].type;
    }
 
  /* Merge Tag_compatibility attributes and any common GNU ones.  */
  if (!_bfd_elf_merge_object_attributes (ibfd, obfd))
    return FALSE;
 
  /* Check for any attributes not known on ARM.  */
  result &= _bfd_elf_merge_unknown_attribute_list (ibfd, obfd);
 
  return result;
}
 
 
/* Return TRUE if the two EABI versions are incompatible.  */
 
static bfd_boolean
elf32_arm_versions_compatible (unsigned iver, unsigned over)
{
  /* v4 and v5 are the same spec before and after it was released,
     so allow mixing them.  */
  if ((iver == EF_ARM_EABI_VER4 && over == EF_ARM_EABI_VER5)
      || (iver == EF_ARM_EABI_VER5 && over == EF_ARM_EABI_VER4))
    return TRUE;
 
  return (iver == over);
}
 
/* Merge backend specific data from an object file to the output
   object file when linking.  */
 
static bfd_boolean
elf32_arm_merge_private_bfd_data (bfd * ibfd, bfd * obfd);
 
/* Display the flags field.  */
 
static bfd_boolean
elf32_arm_print_private_bfd_data (bfd *abfd, void * ptr)
{
  FILE * file = (FILE *) ptr;
  unsigned long flags;
 
  BFD_ASSERT (abfd != NULL && ptr != NULL);
 
  /* Print normal ELF private data.  */
  _bfd_elf_print_private_bfd_data (abfd, ptr);
 
  flags = elf_elfheader (abfd)->e_flags;
  /* Ignore init flag - it may not be set, despite the flags field
     containing valid data.  */
 
  /* xgettext:c-format */
  fprintf (file, _("private flags = %lx:"), elf_elfheader (abfd)->e_flags);
 
  switch (EF_ARM_EABI_VERSION (flags))
    {
    case EF_ARM_EABI_UNKNOWN:
      /* The following flag bits are GNU extensions and not part of the
	 official ARM ELF extended ABI.  Hence they are only decoded if
	 the EABI version is not set.  */
      if (flags & EF_ARM_INTERWORK)
	fprintf (file, _(" [interworking enabled]"));
 
      if (flags & EF_ARM_APCS_26)
	fprintf (file, " [APCS-26]");
      else
	fprintf (file, " [APCS-32]");
 
      if (flags & EF_ARM_VFP_FLOAT)
	fprintf (file, _(" [VFP float format]"));
      else if (flags & EF_ARM_MAVERICK_FLOAT)
	fprintf (file, _(" [Maverick float format]"));
      else
	fprintf (file, _(" [FPA float format]"));
 
      if (flags & EF_ARM_APCS_FLOAT)
	fprintf (file, _(" [floats passed in float registers]"));
 
      if (flags & EF_ARM_PIC)
	fprintf (file, _(" [position independent]"));
 
      if (flags & EF_ARM_NEW_ABI)
	fprintf (file, _(" [new ABI]"));
 
      if (flags & EF_ARM_OLD_ABI)
	fprintf (file, _(" [old ABI]"));
 
      if (flags & EF_ARM_SOFT_FLOAT)
	fprintf (file, _(" [software FP]"));
 
      flags &= ~(EF_ARM_INTERWORK | EF_ARM_APCS_26 | EF_ARM_APCS_FLOAT
		 | EF_ARM_PIC | EF_ARM_NEW_ABI | EF_ARM_OLD_ABI
		 | EF_ARM_SOFT_FLOAT | EF_ARM_VFP_FLOAT
		 | EF_ARM_MAVERICK_FLOAT);
      break;
 
    case EF_ARM_EABI_VER1:
      fprintf (file, _(" [Version1 EABI]"));
 
      if (flags & EF_ARM_SYMSARESORTED)
	fprintf (file, _(" [sorted symbol table]"));
      else
	fprintf (file, _(" [unsorted symbol table]"));
 
      flags &= ~ EF_ARM_SYMSARESORTED;
      break;
 
    case EF_ARM_EABI_VER2:
      fprintf (file, _(" [Version2 EABI]"));
 
      if (flags & EF_ARM_SYMSARESORTED)
	fprintf (file, _(" [sorted symbol table]"));
      else
	fprintf (file, _(" [unsorted symbol table]"));
 
      if (flags & EF_ARM_DYNSYMSUSESEGIDX)
	fprintf (file, _(" [dynamic symbols use segment index]"));
 
      if (flags & EF_ARM_MAPSYMSFIRST)
	fprintf (file, _(" [mapping symbols precede others]"));
 
      flags &= ~(EF_ARM_SYMSARESORTED | EF_ARM_DYNSYMSUSESEGIDX
		 | EF_ARM_MAPSYMSFIRST);
      break;
 
    case EF_ARM_EABI_VER3:
      fprintf (file, _(" [Version3 EABI]"));
      break;
 
    case EF_ARM_EABI_VER4:
      fprintf (file, _(" [Version4 EABI]"));
      goto eabi;
 
    case EF_ARM_EABI_VER5:
      fprintf (file, _(" [Version5 EABI]"));
    eabi:
      if (flags & EF_ARM_BE8)
	fprintf (file, _(" [BE8]"));
 
      if (flags & EF_ARM_LE8)
	fprintf (file, _(" [LE8]"));
 
      flags &= ~(EF_ARM_LE8 | EF_ARM_BE8);
      break;
 
    default:
      fprintf (file, _(" <EABI version unrecognised>"));
      break;
    }
 
  flags &= ~ EF_ARM_EABIMASK;
 
  if (flags & EF_ARM_RELEXEC)
    fprintf (file, _(" [relocatable executable]"));
 
  if (flags & EF_ARM_HASENTRY)
    fprintf (file, _(" [has entry point]"));
 
  flags &= ~ (EF_ARM_RELEXEC | EF_ARM_HASENTRY);
 
  if (flags)
    fprintf (file, _("<Unrecognised flag bits set>"));
 
  fputc ('\n', file);
 
  return TRUE;
}
 
static int
elf32_arm_get_symbol_type (Elf_Internal_Sym * elf_sym, int type)
{
  switch (ELF_ST_TYPE (elf_sym->st_info))
    {
    case STT_ARM_TFUNC:
      return ELF_ST_TYPE (elf_sym->st_info);
 
    case STT_ARM_16BIT:
      /* If the symbol is not an object, return the STT_ARM_16BIT flag.
	 This allows us to distinguish between data used by Thumb instructions
	 and non-data (which is probably code) inside Thumb regions of an
	 executable.  */
      if (type != STT_OBJECT && type != STT_TLS)
	return ELF_ST_TYPE (elf_sym->st_info);
      break;
 
    default:
      break;
    }
 
  return type;
}
 
static asection *
elf32_arm_gc_mark_hook (asection *sec,
			struct bfd_link_info *info,
			Elf_Internal_Rela *rel,
			struct elf_link_hash_entry *h,
			Elf_Internal_Sym *sym)
{
  if (h != NULL)
    switch (ELF32_R_TYPE (rel->r_info))
      {
      case R_ARM_GNU_VTINHERIT:
      case R_ARM_GNU_VTENTRY:
	return NULL;
      }
 
  return _bfd_elf_gc_mark_hook (sec, info, rel, h, sym);
}
 
/* Update the got entry reference counts for the section being removed.  */
 
static bfd_boolean
elf32_arm_gc_sweep_hook (bfd *                     abfd,
			 struct bfd_link_info *    info,
			 asection *                sec,
			 const Elf_Internal_Rela * relocs)
{
  Elf_Internal_Shdr *symtab_hdr;
  struct elf_link_hash_entry **sym_hashes;
  bfd_signed_vma *local_got_refcounts;
  const Elf_Internal_Rela *rel, *relend;
  struct elf32_arm_link_hash_table * globals;
 
  if (info->relocatable)
    return TRUE;
 
  globals = elf32_arm_hash_table (info);
  if (globals == NULL)
    return FALSE;
 
  elf_section_data (sec)->local_dynrel = NULL;
 
  symtab_hdr = & elf_symtab_hdr (abfd);
  sym_hashes = elf_sym_hashes (abfd);
  local_got_refcounts = elf_local_got_refcounts (abfd);
 
  check_use_blx (globals);
 
  relend = relocs + sec->reloc_count;
  for (rel = relocs; rel < relend; rel++)
    {
      unsigned long r_symndx;
      struct elf_link_hash_entry *h = NULL;
      struct elf32_arm_link_hash_entry *eh;
      int r_type;
      bfd_boolean call_reloc_p;
      bfd_boolean may_become_dynamic_p;
      bfd_boolean may_need_local_target_p;
      union gotplt_union *root_plt;
      struct arm_plt_info *arm_plt;
 
      r_symndx = ELF32_R_SYM (rel->r_info);
      if (r_symndx >= symtab_hdr->sh_info)
	{
	  h = sym_hashes[r_symndx - symtab_hdr->sh_info];
	  while (h->root.type == bfd_link_hash_indirect
		 || h->root.type == bfd_link_hash_warning)
	    h = (struct elf_link_hash_entry *) h->root.u.i.link;
	}
      eh = (struct elf32_arm_link_hash_entry *) h;
 
      call_reloc_p = FALSE;
      may_become_dynamic_p = FALSE;
      may_need_local_target_p = FALSE;
 
      r_type = ELF32_R_TYPE (rel->r_info);
      r_type = arm_real_reloc_type (globals, r_type);
      switch (r_type)
	{
	case R_ARM_GOT32:
	case R_ARM_GOT_PREL:
	case R_ARM_TLS_GD32:
	case R_ARM_TLS_IE32:
	  if (h != NULL)
	    {
	      if (h->got.refcount > 0)
		h->got.refcount -= 1;
	    }
	  else if (local_got_refcounts != NULL)
	    {
	      if (local_got_refcounts[r_symndx] > 0)
		local_got_refcounts[r_symndx] -= 1;
	    }
	  break;
 
	case R_ARM_TLS_LDM32:
	  globals->tls_ldm_got.refcount -= 1;
	  break;
 
	case R_ARM_PC24:
	case R_ARM_PLT32:
	case R_ARM_CALL:
	case R_ARM_JUMP24:
	case R_ARM_PREL31:
	case R_ARM_THM_CALL:
	case R_ARM_THM_JUMP24:
	case R_ARM_THM_JUMP19:
	  call_reloc_p = TRUE;
	  may_need_local_target_p = TRUE;
	  break;
 
	case R_ARM_ABS12:
	  if (!globals->vxworks_p)
	    {
	      may_need_local_target_p = TRUE;
	      break;
	    }
	  /* Fall through.  */
	case R_ARM_ABS32:
	case R_ARM_ABS32_NOI:
	case R_ARM_REL32:
	case R_ARM_REL32_NOI:
	case R_ARM_MOVW_ABS_NC:
	case R_ARM_MOVT_ABS:
	case R_ARM_MOVW_PREL_NC:
	case R_ARM_MOVT_PREL:
	case R_ARM_THM_MOVW_ABS_NC:
	case R_ARM_THM_MOVT_ABS:
	case R_ARM_THM_MOVW_PREL_NC:
	case R_ARM_THM_MOVT_PREL:
	  /* Should the interworking branches be here also?  */
	  if ((info->shared || globals->root.is_relocatable_executable)
	      && (sec->flags & SEC_ALLOC) != 0)
	    {
	      if (h == NULL
		  && (r_type == R_ARM_REL32 || r_type == R_ARM_REL32_NOI))
		{
		  call_reloc_p = TRUE;
		  may_need_local_target_p = TRUE;
		}
	      else
		may_become_dynamic_p = TRUE;
	    }
	  else
	    may_need_local_target_p = TRUE;
	  break;
 
	default:
	  break;
	}
 
      if (may_need_local_target_p
	  && elf32_arm_get_plt_info (abfd, eh, r_symndx, &root_plt, &arm_plt))
	{
	  BFD_ASSERT (root_plt->refcount > 0);
	  root_plt->refcount -= 1;
 
	  if (!call_reloc_p)
	    arm_plt->noncall_refcount--;
 
	  if (r_type == R_ARM_THM_CALL)
	    arm_plt->maybe_thumb_refcount--;
 
	  if (r_type == R_ARM_THM_JUMP24
	      || r_type == R_ARM_THM_JUMP19)
	    arm_plt->thumb_refcount--;
	}
 
      if (may_become_dynamic_p)
	{
	  struct elf_dyn_relocs **pp;
	  struct elf_dyn_relocs *p;
 
	  if (h != NULL)
	    pp = &(eh->dyn_relocs);
	  else
	    {
	      Elf_Internal_Sym *isym;
 
	      isym = bfd_sym_from_r_symndx (&globals->sym_cache,
					    abfd, r_symndx);
	      if (isym == NULL)
		return FALSE;
	      pp = elf32_arm_get_local_dynreloc_list (abfd, r_symndx, isym);
	      if (pp == NULL)
		return FALSE;
	    }
	  for (; (p = *pp) != NULL; pp = &p->next)
	    if (p->sec == sec)
	      {
		/* Everything must go for SEC.  */
		*pp = p->next;
		break;
	      }
	}
    }
 
  return TRUE;
}
 
/* Look through the relocs for a section during the first phase.  */
 
static bfd_boolean
elf32_arm_check_relocs (bfd *abfd, struct bfd_link_info *info,
			asection *sec, const Elf_Internal_Rela *relocs)
{
  Elf_Internal_Shdr *symtab_hdr;
  struct elf_link_hash_entry **sym_hashes;
  const Elf_Internal_Rela *rel;
  const Elf_Internal_Rela *rel_end;
  bfd *dynobj;
  asection *sreloc;
  struct elf32_arm_link_hash_table *htab;
  bfd_boolean call_reloc_p;
  bfd_boolean may_become_dynamic_p;
  bfd_boolean may_need_local_target_p;
  unsigned long nsyms;
 
  if (info->relocatable)
    return TRUE;
 
  BFD_ASSERT (is_arm_elf (abfd));
 
  htab = elf32_arm_hash_table (info);
  if (htab == NULL)
    return FALSE;
 
  sreloc = NULL;
 
  /* Create dynamic sections for relocatable executables so that we can
     copy relocations.  */
  if (htab->root.is_relocatable_executable
      && ! htab->root.dynamic_sections_created)
    {
      if (! _bfd_elf_link_create_dynamic_sections (abfd, info))
	return FALSE;
    }
 
  if (htab->root.dynobj == NULL)
    htab->root.dynobj = abfd;
  if (!create_ifunc_sections (info))
    return FALSE;
 
  dynobj = htab->root.dynobj;
 
  symtab_hdr = & elf_symtab_hdr (abfd);
  sym_hashes = elf_sym_hashes (abfd);
  nsyms = NUM_SHDR_ENTRIES (symtab_hdr);
 
  rel_end = relocs + sec->reloc_count;
  for (rel = relocs; rel < rel_end; rel++)
    {
      Elf_Internal_Sym *isym;
      struct elf_link_hash_entry *h;
      struct elf32_arm_link_hash_entry *eh;
      unsigned long r_symndx;
      int r_type;
 
      r_symndx = ELF32_R_SYM (rel->r_info);
      r_type = ELF32_R_TYPE (rel->r_info);
      r_type = arm_real_reloc_type (htab, r_type);
 
      if (r_symndx >= nsyms
	  /* PR 9934: It is possible to have relocations that do not
	     refer to symbols, thus it is also possible to have an
	     object file containing relocations but no symbol table.  */
	  && (r_symndx > STN_UNDEF || nsyms > 0))
	{
	  (*_bfd_error_handler) (_("%B: bad symbol index: %d"), abfd,
				   r_symndx);
	  return FALSE;
	}
 
      h = NULL;
      isym = NULL;
      if (nsyms > 0)
	{
	  if (r_symndx < symtab_hdr->sh_info)
	    {
	      /* A local symbol.  */
	      isym = bfd_sym_from_r_symndx (&htab->sym_cache,
					    abfd, r_symndx);
	      if (isym == NULL)
		return FALSE;
	    }
	  else
	    {
	      h = sym_hashes[r_symndx - symtab_hdr->sh_info];
	      while (h->root.type == bfd_link_hash_indirect
		     || h->root.type == bfd_link_hash_warning)
		h = (struct elf_link_hash_entry *) h->root.u.i.link;
	    }
	}
 
      eh = (struct elf32_arm_link_hash_entry *) h;
 
      call_reloc_p = FALSE;
      may_become_dynamic_p = FALSE;
      may_need_local_target_p = FALSE;
 
      /* Could be done earlier, if h were already available.  */
      r_type = elf32_arm_tls_transition (info, r_type, h);
      switch (r_type)
        {
	  case R_ARM_GOT32:
	  case R_ARM_GOT_PREL:
	  case R_ARM_TLS_GD32:
	  case R_ARM_TLS_IE32:
	  case R_ARM_TLS_GOTDESC:
	  case R_ARM_TLS_DESCSEQ:
	  case R_ARM_THM_TLS_DESCSEQ:
	  case R_ARM_TLS_CALL:
	  case R_ARM_THM_TLS_CALL:
	    /* This symbol requires a global offset table entry.  */
	    {
	      int tls_type, old_tls_type;
 
	      switch (r_type)
		{
		case R_ARM_TLS_GD32: tls_type = GOT_TLS_GD; break;
 
		case R_ARM_TLS_IE32: tls_type = GOT_TLS_IE; break;
 
		case R_ARM_TLS_GOTDESC:
		case R_ARM_TLS_CALL: case R_ARM_THM_TLS_CALL:
		case R_ARM_TLS_DESCSEQ: case R_ARM_THM_TLS_DESCSEQ:
		  tls_type = GOT_TLS_GDESC; break;
 
		default: tls_type = GOT_NORMAL; break;
		}
 
	      if (h != NULL)
		{
		  h->got.refcount++;
		  old_tls_type = elf32_arm_hash_entry (h)->tls_type;
		}
	      else
		{
		  /* This is a global offset table entry for a local symbol.  */
		  if (!elf32_arm_allocate_local_sym_info (abfd))
		    return FALSE;
		  elf_local_got_refcounts (abfd)[r_symndx] += 1;
		  old_tls_type = elf32_arm_local_got_tls_type (abfd) [r_symndx];
		}
 
	      /* If a variable is accessed with both tls methods, two
	         slots may be created.  */
	      if (GOT_TLS_GD_ANY_P (old_tls_type)
		  && GOT_TLS_GD_ANY_P (tls_type))
		tls_type |= old_tls_type;
 
	      /* We will already have issued an error message if there
		 is a TLS/non-TLS mismatch, based on the symbol
		 type.  So just combine any TLS types needed.  */
	      if (old_tls_type != GOT_UNKNOWN && old_tls_type != GOT_NORMAL
		  && tls_type != GOT_NORMAL)
		tls_type |= old_tls_type;
 
	      /* If the symbol is accessed in both IE and GDESC
	         method, we're able to relax. Turn off the GDESC flag,
	         without messing up with any other kind of tls types
	         that may be involved */
	      if ((tls_type & GOT_TLS_IE) && (tls_type & GOT_TLS_GDESC))
		tls_type &= ~GOT_TLS_GDESC;
 
	      if (old_tls_type != tls_type)
		{
		  if (h != NULL)
		    elf32_arm_hash_entry (h)->tls_type = tls_type;
		  else
		    elf32_arm_local_got_tls_type (abfd) [r_symndx] = tls_type;
		}
	    }
	    /* Fall through.  */
 
	  case R_ARM_TLS_LDM32:
	    if (r_type == R_ARM_TLS_LDM32)
		htab->tls_ldm_got.refcount++;
	    /* Fall through.  */
 
	  case R_ARM_GOTOFF32:
	  case R_ARM_GOTPC:
	    if (htab->root.sgot == NULL
		&& !create_got_section (htab->root.dynobj, info))
	      return FALSE;
	    break;
 
	  case R_ARM_PC24:
	  case R_ARM_PLT32:
	  case R_ARM_CALL:
	  case R_ARM_JUMP24:
	  case R_ARM_PREL31:
	  case R_ARM_THM_CALL:
	  case R_ARM_THM_JUMP24:
	  case R_ARM_THM_JUMP19:
	    call_reloc_p = TRUE;
	    may_need_local_target_p = TRUE;
	    break;
 
	  case R_ARM_ABS12:
	    /* VxWorks uses dynamic R_ARM_ABS12 relocations for
	       ldr __GOTT_INDEX__ offsets.  */
	    if (!htab->vxworks_p)
	      {
		may_need_local_target_p = TRUE;
		break;
	      }
	    /* Fall through.  */
 
	  case R_ARM_MOVW_ABS_NC:
	  case R_ARM_MOVT_ABS:
	  case R_ARM_THM_MOVW_ABS_NC:
	  case R_ARM_THM_MOVT_ABS:
	    if (info->shared)
	      {
		(*_bfd_error_handler)
		  (_("%B: relocation %s against `%s' can not be used when making a shared object; recompile with -fPIC"),
		   abfd, elf32_arm_howto_table_1[r_type].name,
		   (h) ? h->root.root.string : "a local symbol");
		bfd_set_error (bfd_error_bad_value);
		return FALSE;
	      }
 
	    /* Fall through.  */
	  case R_ARM_ABS32:
	  case R_ARM_ABS32_NOI:
	  case R_ARM_REL32:
	  case R_ARM_REL32_NOI:
	  case R_ARM_MOVW_PREL_NC:
	  case R_ARM_MOVT_PREL:
	  case R_ARM_THM_MOVW_PREL_NC:
	  case R_ARM_THM_MOVT_PREL:
 
	    /* Should the interworking branches be listed here?  */
	    if ((info->shared || htab->root.is_relocatable_executable)
		&& (sec->flags & SEC_ALLOC) != 0)
	      {
		if (h == NULL
		    && (r_type == R_ARM_REL32 || r_type == R_ARM_REL32_NOI))
		  {
		    /* In shared libraries and relocatable executables,
		       we treat local relative references as calls;
		       see the related SYMBOL_CALLS_LOCAL code in
		       allocate_dynrelocs.  */
		    call_reloc_p = TRUE;
		    may_need_local_target_p = TRUE;
		  }
		else
		  /* We are creating a shared library or relocatable
		     executable, and this is a reloc against a global symbol,
		     or a non-PC-relative reloc against a local symbol.
		     We may need to copy the reloc into the output.  */
		  may_become_dynamic_p = TRUE;
	      }
	    else
	      may_need_local_target_p = TRUE;
	    break;
 
        /* This relocation describes the C++ object vtable hierarchy.
           Reconstruct it for later use during GC.  */
        case R_ARM_GNU_VTINHERIT:
          if (!bfd_elf_gc_record_vtinherit (abfd, sec, h, rel->r_offset))
            return FALSE;
          break;
 
        /* This relocation describes which C++ vtable entries are actually
           used.  Record for later use during GC.  */
        case R_ARM_GNU_VTENTRY:
          BFD_ASSERT (h != NULL);
          if (h != NULL
              && !bfd_elf_gc_record_vtentry (abfd, sec, h, rel->r_offset))
            return FALSE;
          break;
        }
 
      if (h != NULL)
	{
	  if (call_reloc_p)
	    /* We may need a .plt entry if the function this reloc
	       refers to is in a different object, regardless of the
	       symbol's type.  We can't tell for sure yet, because
	       something later might force the symbol local.  */
	    h->needs_plt = 1;
	  else if (may_need_local_target_p)
	    /* If this reloc is in a read-only section, we might
	       need a copy reloc.  We can't check reliably at this
	       stage whether the section is read-only, as input
	       sections have not yet been mapped to output sections.
	       Tentatively set the flag for now, and correct in
	       adjust_dynamic_symbol.  */
	    h->non_got_ref = 1;
	}
 
      if (may_need_local_target_p
	  && (h != NULL || ELF32_ST_TYPE (isym->st_info) == STT_GNU_IFUNC))
	{
	  union gotplt_union *root_plt;
	  struct arm_plt_info *arm_plt;
	  struct arm_local_iplt_info *local_iplt;
 
	  if (h != NULL)
	    {
	      root_plt = &h->plt;
	      arm_plt = &eh->plt;
	    }
	  else
	    {
	      local_iplt = elf32_arm_create_local_iplt (abfd, r_symndx);
	      if (local_iplt == NULL)
		return FALSE;
	      root_plt = &local_iplt->root;
	      arm_plt = &local_iplt->arm;
	    }
 
	  /* If the symbol is a function that doesn't bind locally,
	     this relocation will need a PLT entry.  */
	  root_plt->refcount += 1;
 
	  if (!call_reloc_p)
	    arm_plt->noncall_refcount++;
 
	  /* It's too early to use htab->use_blx here, so we have to
	     record possible blx references separately from
	     relocs that definitely need a thumb stub.  */
 
	  if (r_type == R_ARM_THM_CALL)
	    arm_plt->maybe_thumb_refcount += 1;
 
	  if (r_type == R_ARM_THM_JUMP24
	      || r_type == R_ARM_THM_JUMP19)
	    arm_plt->thumb_refcount += 1;
	}
 
      if (may_become_dynamic_p)
	{
	  struct elf_dyn_relocs *p, **head;
 
	  /* Create a reloc section in dynobj.  */
	  if (sreloc == NULL)
	    {
	      sreloc = _bfd_elf_make_dynamic_reloc_section
		(sec, dynobj, 2, abfd, ! htab->use_rel);
 
	      if (sreloc == NULL)
		return FALSE;
 
	      /* BPABI objects never have dynamic relocations mapped.  */
	      if (htab->symbian_p)
		{
		  flagword flags;
 
		  flags = bfd_get_section_flags (dynobj, sreloc);
		  flags &= ~(SEC_LOAD | SEC_ALLOC);
		  bfd_set_section_flags (dynobj, sreloc, flags);
		}
	    }
 
	  /* If this is a global symbol, count the number of
	     relocations we need for this symbol.  */
	  if (h != NULL)
	    head = &((struct elf32_arm_link_hash_entry *) h)->dyn_relocs;
	  else
	    {
	      head = elf32_arm_get_local_dynreloc_list (abfd, r_symndx, isym);
	      if (head == NULL)
		return FALSE;
	    }
 
	  p = *head;
	  if (p == NULL || p->sec != sec)
	    {
	      bfd_size_type amt = sizeof *p;
 
	      p = (struct elf_dyn_relocs *) bfd_alloc (htab->root.dynobj, amt);
	      if (p == NULL)
		return FALSE;
	      p->next = *head;
	      *head = p;
	      p->sec = sec;
	      p->count = 0;
	      p->pc_count = 0;
	    }
 
	  if (r_type == R_ARM_REL32 || r_type == R_ARM_REL32_NOI)
	    p->pc_count += 1;
	  p->count += 1;
	}
    }
 
  return TRUE;
}
 
/* Unwinding tables are not referenced directly.  This pass marks them as
   required if the corresponding code section is marked.  */
 
static bfd_boolean
elf32_arm_gc_mark_extra_sections (struct bfd_link_info *info,
				  elf_gc_mark_hook_fn gc_mark_hook)
{
  bfd *sub;
  Elf_Internal_Shdr **elf_shdrp;
  bfd_boolean again;
 
  /* Marking EH data may cause additional code sections to be marked,
     requiring multiple passes.  */
  again = TRUE;
  while (again)
    {
      again = FALSE;
      for (sub = info->input_bfds; sub != NULL; sub = sub->link_next)
	{
	  asection *o;
 
	  if (! is_arm_elf (sub))
	    continue;
 
	  elf_shdrp = elf_elfsections (sub);
	  for (o = sub->sections; o != NULL; o = o->next)
	    {
	      Elf_Internal_Shdr *hdr;
 
	      hdr = &elf_section_data (o)->this_hdr;
	      if (hdr->sh_type == SHT_ARM_EXIDX
		  && hdr->sh_link
		  && hdr->sh_link < elf_numsections (sub)
		  && !o->gc_mark
		  && elf_shdrp[hdr->sh_link]->bfd_section->gc_mark)
		{
		  again = TRUE;
		  if (!_bfd_elf_gc_mark (info, o, gc_mark_hook))
		    return FALSE;
		}
	    }
	}
    }
 
  return TRUE;
}
 
/* Treat mapping symbols as special target symbols.  */
 
static bfd_boolean
elf32_arm_is_target_special_symbol (bfd * abfd ATTRIBUTE_UNUSED, asymbol * sym)
{
  return bfd_is_arm_special_symbol_name (sym->name,
					 BFD_ARM_SPECIAL_SYM_TYPE_ANY);
}
 
/* This is a copy of elf_find_function() from elf.c except that
   ARM mapping symbols are ignored when looking for function names
   and STT_ARM_TFUNC is considered to a function type.  */
 
static bfd_boolean
arm_elf_find_function (bfd *         abfd ATTRIBUTE_UNUSED,
		       asection *    section,
		       asymbol **    symbols,
		       bfd_vma       offset,
		       const char ** filename_ptr,
		       const char ** functionname_ptr)
{
  const char * filename = NULL;
  asymbol * func = NULL;
  bfd_vma low_func = 0;
  asymbol ** p;
 
  for (p = symbols; *p != NULL; p++)
    {
      elf_symbol_type *q;
 
      q = (elf_symbol_type *) *p;
 
      switch (ELF_ST_TYPE (q->internal_elf_sym.st_info))
	{
	default:
	  break;
	case STT_FILE:
	  filename = bfd_asymbol_name (&q->symbol);
	  break;
	case STT_FUNC:
	case STT_ARM_TFUNC:
	case STT_NOTYPE:
	  /* Skip mapping symbols.  */
	  if ((q->symbol.flags & BSF_LOCAL)
	      && bfd_is_arm_special_symbol_name (q->symbol.name,
		    BFD_ARM_SPECIAL_SYM_TYPE_ANY))
	    continue;
	  /* Fall through.  */
	  if (bfd_get_section (&q->symbol) == section
	      && q->symbol.value >= low_func
	      && q->symbol.value <= offset)
	    {
	      func = (asymbol *) q;
	      low_func = q->symbol.value;
	    }
	  break;
	}
    }
 
  if (func == NULL)
    return FALSE;
 
  if (filename_ptr)
    *filename_ptr = filename;
  if (functionname_ptr)
    *functionname_ptr = bfd_asymbol_name (func);
 
  return TRUE;
}
 
 
/* Find the nearest line to a particular section and offset, for error
   reporting.   This code is a duplicate of the code in elf.c, except
   that it uses arm_elf_find_function.  */
 
static bfd_boolean
elf32_arm_find_nearest_line (bfd *          abfd,
			     asection *     section,
			     asymbol **     symbols,
			     bfd_vma        offset,
			     const char **  filename_ptr,
			     const char **  functionname_ptr,
			     unsigned int * line_ptr)
{
  bfd_boolean found = FALSE;
 
  /* We skip _bfd_dwarf1_find_nearest_line since no known ARM toolchain uses it.  */
 
  if (_bfd_dwarf2_find_nearest_line (abfd, section, symbols, offset,
				     filename_ptr, functionname_ptr,
				     line_ptr, 0,
				     & elf_tdata (abfd)->dwarf2_find_line_info))
    {
      if (!*functionname_ptr)
	arm_elf_find_function (abfd, section, symbols, offset,
			       *filename_ptr ? NULL : filename_ptr,
			       functionname_ptr);
 
      return TRUE;
    }
 
  if (! _bfd_stab_section_find_nearest_line (abfd, symbols, section, offset,
					     & found, filename_ptr,
					     functionname_ptr, line_ptr,
					     & elf_tdata (abfd)->line_info))
    return FALSE;
 
  if (found && (*functionname_ptr || *line_ptr))
    return TRUE;
 
  if (symbols == NULL)
    return FALSE;
 
  if (! arm_elf_find_function (abfd, section, symbols, offset,
			       filename_ptr, functionname_ptr))
    return FALSE;
 
  *line_ptr = 0;
  return TRUE;
}
 
static bfd_boolean
elf32_arm_find_inliner_info (bfd *          abfd,
			     const char **  filename_ptr,
			     const char **  functionname_ptr,
			     unsigned int * line_ptr)
{
  bfd_boolean found;
  found = _bfd_dwarf2_find_inliner_info (abfd, filename_ptr,
					 functionname_ptr, line_ptr,
					 & elf_tdata (abfd)->dwarf2_find_line_info);
  return found;
}
 
/* Adjust a symbol defined by a dynamic object and referenced by a
   regular object.  The current definition is in some section of the
   dynamic object, but we're not including those sections.  We have to
   change the definition to something the rest of the link can
   understand.  */
 
static bfd_boolean
elf32_arm_adjust_dynamic_symbol (struct bfd_link_info * info,
				 struct elf_link_hash_entry * h)
{
  bfd * dynobj;
  asection * s;
  struct elf32_arm_link_hash_entry * eh;
  struct elf32_arm_link_hash_table *globals;
 
  globals = elf32_arm_hash_table (info);
  if (globals == NULL)
    return FALSE;
 
  dynobj = elf_hash_table (info)->dynobj;
 
  /* Make sure we know what is going on here.  */
  BFD_ASSERT (dynobj != NULL
	      && (h->needs_plt
		  || h->type == STT_GNU_IFUNC
		  || h->u.weakdef != NULL
		  || (h->def_dynamic
		      && h->ref_regular
		      && !h->def_regular)));
 
  eh = (struct elf32_arm_link_hash_entry *) h;
 
  /* If this is a function, put it in the procedure linkage table.  We
     will fill in the contents of the procedure linkage table later,
     when we know the address of the .got section.  */
  if (h->type == STT_FUNC || h->type == STT_GNU_IFUNC || h->needs_plt)
    {
      /* Calls to STT_GNU_IFUNC symbols always use a PLT, even if the
	 symbol binds locally.  */
      if (h->plt.refcount <= 0
	  || (h->type != STT_GNU_IFUNC
	      && (SYMBOL_CALLS_LOCAL (info, h)
		  || (ELF_ST_VISIBILITY (h->other) != STV_DEFAULT
		      && h->root.type == bfd_link_hash_undefweak))))
	{
	  /* This case can occur if we saw a PLT32 reloc in an input
	     file, but the symbol was never referred to by a dynamic
	     object, or if all references were garbage collected.  In
	     such a case, we don't actually need to build a procedure
	     linkage table, and we can just do a PC24 reloc instead.  */
	  h->plt.offset = (bfd_vma) -1;
	  eh->plt.thumb_refcount = 0;
	  eh->plt.maybe_thumb_refcount = 0;
	  eh->plt.noncall_refcount = 0;
	  h->needs_plt = 0;
	}
 
      return TRUE;
    }
  else
    {
      /* It's possible that we incorrectly decided a .plt reloc was
	 needed for an R_ARM_PC24 or similar reloc to a non-function sym
	 in check_relocs.  We can't decide accurately between function
	 and non-function syms in check-relocs; Objects loaded later in
	 the link may change h->type.  So fix it now.  */
      h->plt.offset = (bfd_vma) -1;
      eh->plt.thumb_refcount = 0;
      eh->plt.maybe_thumb_refcount = 0;
      eh->plt.noncall_refcount = 0;
    }
 
  /* If this is a weak symbol, and there is a real definition, the
     processor independent code will have arranged for us to see the
     real definition first, and we can just use the same value.  */
  if (h->u.weakdef != NULL)
    {
      BFD_ASSERT (h->u.weakdef->root.type == bfd_link_hash_defined
		  || h->u.weakdef->root.type == bfd_link_hash_defweak);
      h->root.u.def.section = h->u.weakdef->root.u.def.section;
      h->root.u.def.value = h->u.weakdef->root.u.def.value;
      return TRUE;
    }
 
  /* If there are no non-GOT references, we do not need a copy
     relocation.  */
  if (!h->non_got_ref)
    return TRUE;
 
  /* This is a reference to a symbol defined by a dynamic object which
     is not a function.  */
 
  /* If we are creating a shared library, we must presume that the
     only references to the symbol are via the global offset table.
     For such cases we need not do anything here; the relocations will
     be handled correctly by relocate_section.  Relocatable executables
     can reference data in shared objects directly, so we don't need to
     do anything here.  */
  if (info->shared || globals->root.is_relocatable_executable)
    return TRUE;
 
  if (h->size == 0)
    {
      (*_bfd_error_handler) (_("dynamic variable `%s' is zero size"),
			     h->root.root.string);
      return TRUE;
    }
 
  /* We must allocate the symbol in our .dynbss section, which will
     become part of the .bss section of the executable.  There will be
     an entry for this symbol in the .dynsym section.  The dynamic
     object will contain position independent code, so all references
     from the dynamic object to this symbol will go through the global
     offset table.  The dynamic linker will use the .dynsym entry to
     determine the address it must put in the global offset table, so
     both the dynamic object and the regular object will refer to the
     same memory location for the variable.  */
  s = bfd_get_section_by_name (dynobj, ".dynbss");
  BFD_ASSERT (s != NULL);
 
  /* We must generate a R_ARM_COPY reloc to tell the dynamic linker to
     copy the initial value out of the dynamic object and into the
     runtime process image.  We need to remember the offset into the
     .rel(a).bss section we are going to use.  */
  if ((h->root.u.def.section->flags & SEC_ALLOC) != 0)
    {
      asection *srel;
 
      srel = bfd_get_section_by_name (dynobj, RELOC_SECTION (globals, ".bss"));
      elf32_arm_allocate_dynrelocs (info, srel, 1);
      h->needs_copy = 1;
    }
 
  return _bfd_elf_adjust_dynamic_copy (h, s);
}
 
/* Allocate space in .plt, .got and associated reloc sections for
   dynamic relocs.  */
 
static bfd_boolean
allocate_dynrelocs_for_symbol (struct elf_link_hash_entry *h, void * inf)
{
  struct bfd_link_info *info;
  struct elf32_arm_link_hash_table *htab;
  struct elf32_arm_link_hash_entry *eh;
  struct elf_dyn_relocs *p;
 
  if (h->root.type == bfd_link_hash_indirect)
    return TRUE;
 
  if (h->root.type == bfd_link_hash_warning)
    /* When warning symbols are created, they **replace** the "real"
       entry in the hash table, thus we never get to see the real
       symbol in a hash traversal.  So look at it now.  */
    h = (struct elf_link_hash_entry *) h->root.u.i.link;
 
  eh = (struct elf32_arm_link_hash_entry *) h;
 
  info = (struct bfd_link_info *) inf;
  htab = elf32_arm_hash_table (info);
  if (htab == NULL)
    return FALSE;
 
  if ((htab->root.dynamic_sections_created || h->type == STT_GNU_IFUNC)
      && h->plt.refcount > 0)
    {
      /* Make sure this symbol is output as a dynamic symbol.
	 Undefined weak syms won't yet be marked as dynamic.  */
      if (h->dynindx == -1
	  && !h->forced_local)
	{
	  if (! bfd_elf_link_record_dynamic_symbol (info, h))
	    return FALSE;
	}
 
      /* If the call in the PLT entry binds locally, the associated
	 GOT entry should use an R_ARM_IRELATIVE relocation instead of
	 the usual R_ARM_JUMP_SLOT.  Put it in the .iplt section rather
	 than the .plt section.  */
      if (h->type == STT_GNU_IFUNC && SYMBOL_CALLS_LOCAL (info, h))
	{
	  eh->is_iplt = 1;
	  if (eh->plt.noncall_refcount == 0
	      && SYMBOL_REFERENCES_LOCAL (info, h))
	    /* All non-call references can be resolved directly.
	       This means that they can (and in some cases, must)
	       resolve directly to the run-time target, rather than
	       to the PLT.  That in turns means that any .got entry
	       would be equal to the .igot.plt entry, so there's
	       no point having both.  */
	    h->got.refcount = 0;
	}
 
      if (info->shared
	  || eh->is_iplt
	  || WILL_CALL_FINISH_DYNAMIC_SYMBOL (1, 0, h))
	{
	  elf32_arm_allocate_plt_entry (info, eh->is_iplt, &h->plt, &eh->plt);
 
	  /* If this symbol is not defined in a regular file, and we are
	     not generating a shared library, then set the symbol to this
	     location in the .plt.  This is required to make function
	     pointers compare as equal between the normal executable and
	     the shared library.  */
	  if (! info->shared
	      && !h->def_regular)
	    {
	      h->root.u.def.section = htab->root.splt;
	      h->root.u.def.value = h->plt.offset;
 
	      /* Make sure the function is not marked as Thumb, in case
		 it is the target of an ABS32 relocation, which will
		 point to the PLT entry.  */
	      h->target_internal = ST_BRANCH_TO_ARM;
	    }
 
	  htab->next_tls_desc_index++;
 
	  /* VxWorks executables have a second set of relocations for
	     each PLT entry.  They go in a separate relocation section,
	     which is processed by the kernel loader.  */
	  if (htab->vxworks_p && !info->shared)
	    {
	      /* There is a relocation for the initial PLT entry:
		 an R_ARM_32 relocation for _GLOBAL_OFFSET_TABLE_.  */
	      if (h->plt.offset == htab->plt_header_size)
		elf32_arm_allocate_dynrelocs (info, htab->srelplt2, 1);
 
	      /* There are two extra relocations for each subsequent
		 PLT entry: an R_ARM_32 relocation for the GOT entry,
		 and an R_ARM_32 relocation for the PLT entry.  */
	      elf32_arm_allocate_dynrelocs (info, htab->srelplt2, 2);
	    }
	}
      else
	{
	  h->plt.offset = (bfd_vma) -1;
	  h->needs_plt = 0;
	}
    }
  else
    {
      h->plt.offset = (bfd_vma) -1;
      h->needs_plt = 0;
    }
 
  eh = (struct elf32_arm_link_hash_entry *) h;
  eh->tlsdesc_got = (bfd_vma) -1;
 
  if (h->got.refcount > 0)
    {
      asection *s;
      bfd_boolean dyn;
      int tls_type = elf32_arm_hash_entry (h)->tls_type;
      int indx;
 
      /* Make sure this symbol is output as a dynamic symbol.
	 Undefined weak syms won't yet be marked as dynamic.  */
      if (h->dynindx == -1
	  && !h->forced_local)
	{
	  if (! bfd_elf_link_record_dynamic_symbol (info, h))
	    return FALSE;
	}
 
      if (!htab->symbian_p)
	{
	  s = htab->root.sgot;
	  h->got.offset = s->size;
 
	  if (tls_type == GOT_UNKNOWN)
	    abort ();
 
	  if (tls_type == GOT_NORMAL)
	    /* Non-TLS symbols need one GOT slot.  */
	    s->size += 4;
	  else
	    {
              if (tls_type & GOT_TLS_GDESC)
	        {
		  /* R_ARM_TLS_DESC needs 2 GOT slots.  */
	          eh->tlsdesc_got
		    = (htab->root.sgotplt->size
		       - elf32_arm_compute_jump_table_size (htab));
	          htab->root.sgotplt->size += 8;
	          h->got.offset = (bfd_vma) -2;
		  /* plt.got_offset needs to know there's a TLS_DESC
		     reloc in the middle of .got.plt.  */
                  htab->num_tls_desc++;
	        }
 
	      if (tls_type & GOT_TLS_GD)
		{
		  /* R_ARM_TLS_GD32 needs 2 consecutive GOT slots.  If
		     the symbol is both GD and GDESC, got.offset may
		     have been overwritten.  */
		  h->got.offset = s->size;
		  s->size += 8;
		}
 
	      if (tls_type & GOT_TLS_IE)
		/* R_ARM_TLS_IE32 needs one GOT slot.  */
		s->size += 4;
	    }
 
	  dyn = htab->root.dynamic_sections_created;
 
	  indx = 0;
	  if (WILL_CALL_FINISH_DYNAMIC_SYMBOL (dyn, info->shared, h)
	      && (!info->shared
		  || !SYMBOL_REFERENCES_LOCAL (info, h)))
	    indx = h->dynindx;
 
	  if (tls_type != GOT_NORMAL
	      && (info->shared || indx != 0)
	      && (ELF_ST_VISIBILITY (h->other) == STV_DEFAULT
		  || h->root.type != bfd_link_hash_undefweak))
	    {
	      if (tls_type & GOT_TLS_IE)
		elf32_arm_allocate_dynrelocs (info, htab->root.srelgot, 1);
 
	      if (tls_type & GOT_TLS_GD)
		elf32_arm_allocate_dynrelocs (info, htab->root.srelgot, 1);
 
	      if (tls_type & GOT_TLS_GDESC) 
		{
		  elf32_arm_allocate_dynrelocs (info, htab->root.srelplt, 1);
		  /* GDESC needs a trampoline to jump to.  */
		  htab->tls_trampoline = -1;
		}
 
	      /* Only GD needs it.  GDESC just emits one relocation per
		 2 entries.  */
	      if ((tls_type & GOT_TLS_GD) && indx != 0)  
		elf32_arm_allocate_dynrelocs (info, htab->root.srelgot, 1);
	    }
	  else if (!SYMBOL_REFERENCES_LOCAL (info, h))
	    {
	      if (htab->root.dynamic_sections_created)
		/* Reserve room for the GOT entry's R_ARM_GLOB_DAT relocation.  */
		elf32_arm_allocate_dynrelocs (info, htab->root.srelgot, 1);
	    }
	  else if (h->type == STT_GNU_IFUNC
		   && eh->plt.noncall_refcount == 0)
	    /* No non-call references resolve the STT_GNU_IFUNC's PLT entry;
	       they all resolve dynamically instead.  Reserve room for the
	       GOT entry's R_ARM_IRELATIVE relocation.  */
	    elf32_arm_allocate_irelocs (info, htab->root.srelgot, 1);
	  else if (info->shared)
	    /* Reserve room for the GOT entry's R_ARM_RELATIVE relocation.  */
	    elf32_arm_allocate_dynrelocs (info, htab->root.srelgot, 1);
	}
    }
  else
    h->got.offset = (bfd_vma) -1;
 
  /* Allocate stubs for exported Thumb functions on v4t.  */
  if (!htab->use_blx && h->dynindx != -1
      && h->def_regular
      && h->target_internal == ST_BRANCH_TO_THUMB
      && ELF_ST_VISIBILITY (h->other) == STV_DEFAULT)
    {
      struct elf_link_hash_entry * th;
      struct bfd_link_hash_entry * bh;
      struct elf_link_hash_entry * myh;
      char name[1024];
      asection *s;
      bh = NULL;
      /* Create a new symbol to regist the real location of the function.  */
      s = h->root.u.def.section;
      sprintf (name, "__real_%s", h->root.root.string);
      _bfd_generic_link_add_one_symbol (info, s->owner,
					name, BSF_GLOBAL, s,
					h->root.u.def.value,
					NULL, TRUE, FALSE, &bh);
 
      myh = (struct elf_link_hash_entry *) bh;
      myh->type = ELF_ST_INFO (STB_LOCAL, STT_FUNC);
      myh->forced_local = 1;
      myh->target_internal = ST_BRANCH_TO_THUMB;
      eh->export_glue = myh;
      th = record_arm_to_thumb_glue (info, h);
      /* Point the symbol at the stub.  */
      h->type = ELF_ST_INFO (ELF_ST_BIND (h->type), STT_FUNC);
      h->target_internal = ST_BRANCH_TO_ARM;
      h->root.u.def.section = th->root.u.def.section;
      h->root.u.def.value = th->root.u.def.value & ~1;
    }
 
  if (eh->dyn_relocs == NULL)
    return TRUE;
 
  /* In the shared -Bsymbolic case, discard space allocated for
     dynamic pc-relative relocs against symbols which turn out to be
     defined in regular objects.  For the normal shared case, discard
     space for pc-relative relocs that have become local due to symbol
     visibility changes.  */
 
  if (info->shared || htab->root.is_relocatable_executable)
    {
      /* The only relocs that use pc_count are R_ARM_REL32 and
         R_ARM_REL32_NOI, which will appear on something like
         ".long foo - .".  We want calls to protected symbols to resolve
         directly to the function rather than going via the plt.  If people
         want function pointer comparisons to work as expected then they
         should avoid writing assembly like ".long foo - .".  */
      if (SYMBOL_CALLS_LOCAL (info, h))
	{
	  struct elf_dyn_relocs **pp;
 
	  for (pp = &eh->dyn_relocs; (p = *pp) != NULL; )
	    {
	      p->count -= p->pc_count;
	      p->pc_count = 0;
	      if (p->count == 0)
		*pp = p->next;
	      else
		pp = &p->next;
	    }
	}
 
      if (htab->vxworks_p)
	{
	  struct elf_dyn_relocs **pp;
 
	  for (pp = &eh->dyn_relocs; (p = *pp) != NULL; )
	    {
	      if (strcmp (p->sec->output_section->name, ".tls_vars") == 0)
		*pp = p->next;
	      else
		pp = &p->next;
	    }
	}
 
      /* Also discard relocs on undefined weak syms with non-default
         visibility.  */
      if (eh->dyn_relocs != NULL
	  && h->root.type == bfd_link_hash_undefweak)
	{
	  if (ELF_ST_VISIBILITY (h->other) != STV_DEFAULT)
	    eh->dyn_relocs = NULL;
 
	  /* Make sure undefined weak symbols are output as a dynamic
	     symbol in PIEs.  */
	  else if (h->dynindx == -1
		   && !h->forced_local)
	    {
	      if (! bfd_elf_link_record_dynamic_symbol (info, h))
		return FALSE;
	    }
	}
 
      else if (htab->root.is_relocatable_executable && h->dynindx == -1
	       && h->root.type == bfd_link_hash_new)
	{
	  /* Output absolute symbols so that we can create relocations
	     against them.  For normal symbols we output a relocation
	     against the section that contains them.  */
	  if (! bfd_elf_link_record_dynamic_symbol (info, h))
	    return FALSE;
	}
 
    }
  else
    {
      /* For the non-shared case, discard space for relocs against
	 symbols which turn out to need copy relocs or are not
	 dynamic.  */
 
      if (!h->non_got_ref
	  && ((h->def_dynamic
	       && !h->def_regular)
	      || (htab->root.dynamic_sections_created
		  && (h->root.type == bfd_link_hash_undefweak
		      || h->root.type == bfd_link_hash_undefined))))
	{
	  /* Make sure this symbol is output as a dynamic symbol.
	     Undefined weak syms won't yet be marked as dynamic.  */
	  if (h->dynindx == -1
	      && !h->forced_local)
	    {
	      if (! bfd_elf_link_record_dynamic_symbol (info, h))
		return FALSE;
	    }
 
	  /* If that succeeded, we know we'll be keeping all the
	     relocs.  */
	  if (h->dynindx != -1)
	    goto keep;
	}
 
      eh->dyn_relocs = NULL;
 
    keep: ;
    }
 
  /* Finally, allocate space.  */
  for (p = eh->dyn_relocs; p != NULL; p = p->next)
    {
      asection *sreloc = elf_section_data (p->sec)->sreloc;
      if (h->type == STT_GNU_IFUNC
	  && eh->plt.noncall_refcount == 0
	  && SYMBOL_REFERENCES_LOCAL (info, h))
	elf32_arm_allocate_irelocs (info, sreloc, p->count);
      else
	elf32_arm_allocate_dynrelocs (info, sreloc, p->count);
    }
 
  return TRUE;
}
 
/* Find any dynamic relocs that apply to read-only sections.  */
 
static bfd_boolean
elf32_arm_readonly_dynrelocs (struct elf_link_hash_entry * h, void * inf)
{
  struct elf32_arm_link_hash_entry * eh;
  struct elf_dyn_relocs * p;
 
  if (h->root.type == bfd_link_hash_warning)
    h = (struct elf_link_hash_entry *) h->root.u.i.link;
 
  eh = (struct elf32_arm_link_hash_entry *) h;
  for (p = eh->dyn_relocs; p != NULL; p = p->next)
    {
      asection *s = p->sec;
 
      if (s != NULL && (s->flags & SEC_READONLY) != 0)
	{
	  struct bfd_link_info *info = (struct bfd_link_info *) inf;
 
	  info->flags |= DF_TEXTREL;
 
	  /* Not an error, just cut short the traversal.  */
	  return FALSE;
	}
    }
  return TRUE;
}
 
void
bfd_elf32_arm_set_byteswap_code (struct bfd_link_info *info,
				 int byteswap_code)
{
  struct elf32_arm_link_hash_table *globals;
 
  globals = elf32_arm_hash_table (info);
  if (globals == NULL)
    return;
 
  globals->byteswap_code = byteswap_code;
}
 
/* Set the sizes of the dynamic sections.  */
 
static bfd_boolean
elf32_arm_size_dynamic_sections (bfd * output_bfd ATTRIBUTE_UNUSED,
				 struct bfd_link_info * info)
{
  bfd * dynobj;
  asection * s;
  bfd_boolean plt;
  bfd_boolean relocs;
  bfd *ibfd;
  struct elf32_arm_link_hash_table *htab;
 
  htab = elf32_arm_hash_table (info);
  if (htab == NULL)
    return FALSE;
 
  dynobj = elf_hash_table (info)->dynobj;
  BFD_ASSERT (dynobj != NULL);
  check_use_blx (htab);
 
  if (elf_hash_table (info)->dynamic_sections_created)
    {
      /* Set the contents of the .interp section to the interpreter.  */
      if (info->executable)
	{
	  s = bfd_get_section_by_name (dynobj, ".interp");
	  BFD_ASSERT (s != NULL);
	  s->size = sizeof ELF_DYNAMIC_INTERPRETER;
	  s->contents = (unsigned char *) ELF_DYNAMIC_INTERPRETER;
	}
    }
 
  /* Set up .got offsets for local syms, and space for local dynamic
     relocs.  */
  for (ibfd = info->input_bfds; ibfd != NULL; ibfd = ibfd->link_next)
    {
      bfd_signed_vma *local_got;
      bfd_signed_vma *end_local_got;
      struct arm_local_iplt_info **local_iplt_ptr, *local_iplt;
      char *local_tls_type;
      bfd_vma *local_tlsdesc_gotent;
      bfd_size_type locsymcount;
      Elf_Internal_Shdr *symtab_hdr;
      asection *srel;
      bfd_boolean is_vxworks = htab->vxworks_p;
      unsigned int symndx;
 
      if (! is_arm_elf (ibfd))
	continue;
 
      for (s = ibfd->sections; s != NULL; s = s->next)
	{
	  struct elf_dyn_relocs *p;
 
	  for (p = (struct elf_dyn_relocs *)
                   elf_section_data (s)->local_dynrel; p != NULL; p = p->next)
	    {
	      if (!bfd_is_abs_section (p->sec)
		  && bfd_is_abs_section (p->sec->output_section))
		{
		  /* Input section has been discarded, either because
		     it is a copy of a linkonce section or due to
		     linker script /DISCARD/, so we'll be discarding
		     the relocs too.  */
		}
	      else if (is_vxworks
		       && strcmp (p->sec->output_section->name,
				  ".tls_vars") == 0)
		{
		  /* Relocations in vxworks .tls_vars sections are
		     handled specially by the loader.  */
		}
	      else if (p->count != 0)
		{
		  srel = elf_section_data (p->sec)->sreloc;
		  elf32_arm_allocate_dynrelocs (info, srel, p->count);
		  if ((p->sec->output_section->flags & SEC_READONLY) != 0)
		    info->flags |= DF_TEXTREL;
		}
	    }
	}
 
      local_got = elf_local_got_refcounts (ibfd);
      if (!local_got)
	continue;
 
      symtab_hdr = & elf_symtab_hdr (ibfd);
      locsymcount = symtab_hdr->sh_info;
      end_local_got = local_got + locsymcount;
      local_iplt_ptr = elf32_arm_local_iplt (ibfd);
      local_tls_type = elf32_arm_local_got_tls_type (ibfd);
      local_tlsdesc_gotent = elf32_arm_local_tlsdesc_gotent (ibfd);
      symndx = 0;
      s = htab->root.sgot;
      srel = htab->root.srelgot;
      for (; local_got < end_local_got;
	   ++local_got, ++local_iplt_ptr, ++local_tls_type,
	   ++local_tlsdesc_gotent, ++symndx)
	{
	  *local_tlsdesc_gotent = (bfd_vma) -1;
	  local_iplt = *local_iplt_ptr;
	  if (local_iplt != NULL)
	    {
	      struct elf_dyn_relocs *p;
 
	      if (local_iplt->root.refcount > 0)
		{
		  elf32_arm_allocate_plt_entry (info, TRUE,
						&local_iplt->root,
						&local_iplt->arm);
		  if (local_iplt->arm.noncall_refcount == 0)
		    /* All references to the PLT are calls, so all
		       non-call references can resolve directly to the
		       run-time target.  This means that the .got entry
		       would be the same as the .igot.plt entry, so there's
		       no point creating both.  */
		    *local_got = 0;
		}
	      else
		{
		  BFD_ASSERT (local_iplt->arm.noncall_refcount == 0);
		  local_iplt->root.offset = (bfd_vma) -1;
		}
 
	      for (p = local_iplt->dyn_relocs; p != NULL; p = p->next)
		{
		  asection *psrel;
 
		  psrel = elf_section_data (p->sec)->sreloc;
		  if (local_iplt->arm.noncall_refcount == 0)
		    elf32_arm_allocate_irelocs (info, psrel, p->count);
		  else
		    elf32_arm_allocate_dynrelocs (info, psrel, p->count);
		}
	    }
	  if (*local_got > 0)
	    {
	      Elf_Internal_Sym *isym;
 
	      *local_got = s->size;
	      if (*local_tls_type & GOT_TLS_GD)
		/* TLS_GD relocs need an 8-byte structure in the GOT.  */
		s->size += 8;
	      if (*local_tls_type & GOT_TLS_GDESC)
		{
		  *local_tlsdesc_gotent = htab->root.sgotplt->size
		    - elf32_arm_compute_jump_table_size (htab);
		  htab->root.sgotplt->size += 8;
		  *local_got = (bfd_vma) -2;
		  /* plt.got_offset needs to know there's a TLS_DESC
		     reloc in the middle of .got.plt.  */
                  htab->num_tls_desc++;
		}
	      if (*local_tls_type & GOT_TLS_IE)
		s->size += 4;
 
	      if (*local_tls_type & GOT_NORMAL)
		{
		  /* If the symbol is both GD and GDESC, *local_got
		     may have been overwritten.  */
		  *local_got = s->size;
		  s->size += 4;
		}
 
	      isym = bfd_sym_from_r_symndx (&htab->sym_cache, ibfd, symndx);
	      if (isym == NULL)
		return FALSE;
 
	      /* If all references to an STT_GNU_IFUNC PLT are calls,
		 then all non-call references, including this GOT entry,
		 resolve directly to the run-time target.  */
	      if (ELF32_ST_TYPE (isym->st_info) == STT_GNU_IFUNC
		  && (local_iplt == NULL
		      || local_iplt->arm.noncall_refcount == 0))
		elf32_arm_allocate_irelocs (info, srel, 1);
	      else if ((info->shared && !(*local_tls_type & GOT_TLS_GDESC))
		       || *local_tls_type & GOT_TLS_GD)
		elf32_arm_allocate_dynrelocs (info, srel, 1);
 
	      if (info->shared && *local_tls_type & GOT_TLS_GDESC)
		{
		  elf32_arm_allocate_dynrelocs (info, htab->root.srelplt, 1);
		  htab->tls_trampoline = -1;
		}
	    }
	  else
	    *local_got = (bfd_vma) -1;
	}
    }
 
  if (htab->tls_ldm_got.refcount > 0)
    {
      /* Allocate two GOT entries and one dynamic relocation (if necessary)
	 for R_ARM_TLS_LDM32 relocations.  */
      htab->tls_ldm_got.offset = htab->root.sgot->size;
      htab->root.sgot->size += 8;
      if (info->shared)
	elf32_arm_allocate_dynrelocs (info, htab->root.srelgot, 1);
    }
  else
    htab->tls_ldm_got.offset = -1;
 
  /* Allocate global sym .plt and .got entries, and space for global
     sym dynamic relocs.  */
  elf_link_hash_traverse (& htab->root, allocate_dynrelocs_for_symbol, info);
 
  /* Here we rummage through the found bfds to collect glue information.  */
  for (ibfd = info->input_bfds; ibfd != NULL; ibfd = ibfd->link_next)
    {
      if (! is_arm_elf (ibfd))
	continue;
 
      /* Initialise mapping tables for code/data.  */
      bfd_elf32_arm_init_maps (ibfd);
 
      if (!bfd_elf32_arm_process_before_allocation (ibfd, info)
	  || !bfd_elf32_arm_vfp11_erratum_scan (ibfd, info))
        /* xgettext:c-format */
        _bfd_error_handler (_("Errors encountered processing file %s"),
			    ibfd->filename);
    }
 
  /* Allocate space for the glue sections now that we've sized them.  */
  bfd_elf32_arm_allocate_interworking_sections (info);
 
  /* For every jump slot reserved in the sgotplt, reloc_count is
     incremented.  However, when we reserve space for TLS descriptors,
     it's not incremented, so in order to compute the space reserved
     for them, it suffices to multiply the reloc count by the jump
     slot size.  */
  if (htab->root.srelplt)
    htab->sgotplt_jump_table_size = elf32_arm_compute_jump_table_size(htab);
 
  if (htab->tls_trampoline)
    {
      if (htab->root.splt->size == 0)
	htab->root.splt->size += htab->plt_header_size;
 
      htab->tls_trampoline = htab->root.splt->size;
      htab->root.splt->size += htab->plt_entry_size;
 
      /* If we're not using lazy TLS relocations, don't generate the
         PLT and GOT entries they require.  */
      if (!(info->flags & DF_BIND_NOW))
	{
	  htab->dt_tlsdesc_got = htab->root.sgot->size;
	  htab->root.sgot->size += 4;
 
	  htab->dt_tlsdesc_plt = htab->root.splt->size;
	  htab->root.splt->size += 4 * ARRAY_SIZE (dl_tlsdesc_lazy_trampoline);
	}
    }
 
  /* The check_relocs and adjust_dynamic_symbol entry points have
     determined the sizes of the various dynamic sections.  Allocate
     memory for them.  */
  plt = FALSE;
  relocs = FALSE;
  for (s = dynobj->sections; s != NULL; s = s->next)
    {
      const char * name;
 
      if ((s->flags & SEC_LINKER_CREATED) == 0)
	continue;
 
      /* It's OK to base decisions on the section name, because none
	 of the dynobj section names depend upon the input files.  */
      name = bfd_get_section_name (dynobj, s);
 
      if (s == htab->root.splt)
	{
	  /* Remember whether there is a PLT.  */
	  plt = s->size != 0;
	}
      else if (CONST_STRNEQ (name, ".rel"))
	{
	  if (s->size != 0)
	    {
	      /* Remember whether there are any reloc sections other
		 than .rel(a).plt and .rela.plt.unloaded.  */
	      if (s != htab->root.srelplt && s != htab->srelplt2)
		relocs = TRUE;
 
	      /* We use the reloc_count field as a counter if we need
		 to copy relocs into the output file.  */
	      s->reloc_count = 0;
	    }
	}
      else if (s != htab->root.sgot
	       && s != htab->root.sgotplt
	       && s != htab->root.iplt
	       && s != htab->root.igotplt
	       && s != htab->sdynbss)
	{
	  /* It's not one of our sections, so don't allocate space.  */
	  continue;
	}
 
      if (s->size == 0)
	{
	  /* If we don't need this section, strip it from the
	     output file.  This is mostly to handle .rel(a).bss and
	     .rel(a).plt.  We must create both sections in
	     create_dynamic_sections, because they must be created
	     before the linker maps input sections to output
	     sections.  The linker does that before
	     adjust_dynamic_symbol is called, and it is that
	     function which decides whether anything needs to go
	     into these sections.  */
	  s->flags |= SEC_EXCLUDE;
	  continue;
	}
 
      if ((s->flags & SEC_HAS_CONTENTS) == 0)
	continue;
 
      /* Allocate memory for the section contents.  */
      s->contents = (unsigned char *) bfd_zalloc (dynobj, s->size);
      if (s->contents == NULL)
	return FALSE;
    }
 
  if (elf_hash_table (info)->dynamic_sections_created)
    {
      /* Add some entries to the .dynamic section.  We fill in the
	 values later, in elf32_arm_finish_dynamic_sections, but we
	 must add the entries now so that we get the correct size for
	 the .dynamic section.  The DT_DEBUG entry is filled in by the
	 dynamic linker and used by the debugger.  */
#define add_dynamic_entry(TAG, VAL) \
  _bfd_elf_add_dynamic_entry (info, TAG, VAL)
 
     if (info->executable)
	{
	  if (!add_dynamic_entry (DT_DEBUG, 0))
	    return FALSE;
	}
 
      if (plt)
	{
	  if (   !add_dynamic_entry (DT_PLTGOT, 0)
	      || !add_dynamic_entry (DT_PLTRELSZ, 0)
	      || !add_dynamic_entry (DT_PLTREL,
				     htab->use_rel ? DT_REL : DT_RELA)
	      || !add_dynamic_entry (DT_JMPREL, 0))
	    return FALSE;
 
	  if (htab->dt_tlsdesc_plt &&
		(!add_dynamic_entry (DT_TLSDESC_PLT,0) 
		 || !add_dynamic_entry (DT_TLSDESC_GOT,0)))
	    return FALSE; 
	}
 
      if (relocs)
	{
	  if (htab->use_rel)
	    {
	      if (!add_dynamic_entry (DT_REL, 0)
		  || !add_dynamic_entry (DT_RELSZ, 0)
		  || !add_dynamic_entry (DT_RELENT, RELOC_SIZE (htab)))
		return FALSE;
	    }
	  else
	    {
	      if (!add_dynamic_entry (DT_RELA, 0)
		  || !add_dynamic_entry (DT_RELASZ, 0)
		  || !add_dynamic_entry (DT_RELAENT, RELOC_SIZE (htab)))
		return FALSE;
	    }
	}
 
      /* If any dynamic relocs apply to a read-only section,
	 then we need a DT_TEXTREL entry.  */
      if ((info->flags & DF_TEXTREL) == 0)
	elf_link_hash_traverse (& htab->root, elf32_arm_readonly_dynrelocs,
				info);
 
      if ((info->flags & DF_TEXTREL) != 0)
	{
	  if (!add_dynamic_entry (DT_TEXTREL, 0))
	    return FALSE;
	}
      if (htab->vxworks_p
	  && !elf_vxworks_add_dynamic_entries (output_bfd, info))
	return FALSE;
    }
#undef add_dynamic_entry
 
  return TRUE;
}
 
/* Size sections even though they're not dynamic.  We use it to setup
   _TLS_MODULE_BASE_, if needed.  */
 
static bfd_boolean
elf32_arm_always_size_sections (bfd *output_bfd,
	                        struct bfd_link_info *info)
{
  asection *tls_sec;
 
  if (info->relocatable)
    return TRUE;
 
  tls_sec = elf_hash_table (info)->tls_sec;
 
  if (tls_sec)
    {
      struct elf_link_hash_entry *tlsbase;
 
      tlsbase = elf_link_hash_lookup
	(elf_hash_table (info), "_TLS_MODULE_BASE_", TRUE, TRUE, FALSE);
 
      if (tlsbase)
        {
          struct bfd_link_hash_entry *bh = NULL;
	  const struct elf_backend_data *bed
            = get_elf_backend_data (output_bfd);
 
          if (!(_bfd_generic_link_add_one_symbol
		(info, output_bfd, "_TLS_MODULE_BASE_", BSF_LOCAL,
		 tls_sec, 0, NULL, FALSE,
		 bed->collect, &bh)))
	    return FALSE;
 
      	  tlsbase->type = STT_TLS;
          tlsbase = (struct elf_link_hash_entry *)bh;
          tlsbase->def_regular = 1;
          tlsbase->other = STV_HIDDEN;
          (*bed->elf_backend_hide_symbol) (info, tlsbase, TRUE);
	}
    }
  return TRUE;
}
 
/* Finish up dynamic symbol handling.  We set the contents of various
   dynamic sections here.  */
 
static bfd_boolean
elf32_arm_finish_dynamic_symbol (bfd * output_bfd,
				 struct bfd_link_info * info,
				 struct elf_link_hash_entry * h,
				 Elf_Internal_Sym * sym)
{
  struct elf32_arm_link_hash_table *htab;
  struct elf32_arm_link_hash_entry *eh;
 
  htab = elf32_arm_hash_table (info);
  if (htab == NULL)
    return FALSE;
 
  eh = (struct elf32_arm_link_hash_entry *) h;
 
  if (h->plt.offset != (bfd_vma) -1)
    {
      if (!eh->is_iplt)
	{
	  BFD_ASSERT (h->dynindx != -1);
	  elf32_arm_populate_plt_entry (output_bfd, info, &h->plt, &eh->plt,
					h->dynindx, 0);
	}
 
      if (!h->def_regular)
	{
	  /* Mark the symbol as undefined, rather than as defined in
	     the .plt section.  Leave the value alone.  */
	  sym->st_shndx = SHN_UNDEF;
	  /* If the symbol is weak, we do need to clear the value.
	     Otherwise, the PLT entry would provide a definition for
	     the symbol even if the symbol wasn't defined anywhere,
	     and so the symbol would never be NULL.  */
	  if (!h->ref_regular_nonweak)
	    sym->st_value = 0;
	}
      else if (eh->is_iplt && eh->plt.noncall_refcount != 0)
	{
	  /* At least one non-call relocation references this .iplt entry,
	     so the .iplt entry is the function's canonical address.  */
	  sym->st_info = ELF_ST_INFO (ELF_ST_BIND (sym->st_info), STT_FUNC);
	  sym->st_target_internal = ST_BRANCH_TO_ARM;
	  sym->st_shndx = (_bfd_elf_section_from_bfd_section
			   (output_bfd, htab->root.iplt->output_section));
	  sym->st_value = (h->plt.offset
			   + htab->root.iplt->output_section->vma
			   + htab->root.iplt->output_offset);
	}
    }
 
  if (h->needs_copy)
    {
      asection * s;
      Elf_Internal_Rela rel;
 
      /* This symbol needs a copy reloc.  Set it up.  */
      BFD_ASSERT (h->dynindx != -1
		  && (h->root.type == bfd_link_hash_defined
		      || h->root.type == bfd_link_hash_defweak));
 
      s = htab->srelbss;
      BFD_ASSERT (s != NULL);
 
      rel.r_addend = 0;
      rel.r_offset = (h->root.u.def.value
		      + h->root.u.def.section->output_section->vma
		      + h->root.u.def.section->output_offset);
      rel.r_info = ELF32_R_INFO (h->dynindx, R_ARM_COPY);
      elf32_arm_add_dynreloc (output_bfd, info, s, &rel);
    }
 
  /* Mark _DYNAMIC and _GLOBAL_OFFSET_TABLE_ as absolute.  On VxWorks,
     the _GLOBAL_OFFSET_TABLE_ symbol is not absolute: it is relative
     to the ".got" section.  */
  if (strcmp (h->root.root.string, "_DYNAMIC") == 0
      || (!htab->vxworks_p && h == htab->root.hgot))
    sym->st_shndx = SHN_ABS;
 
  return TRUE;
}
 
static void
arm_put_trampoline (struct elf32_arm_link_hash_table *htab, bfd *output_bfd,
		    void *contents,
		    const unsigned long *template, unsigned count)
{
  unsigned ix;
 
  for (ix = 0; ix != count; ix++)
    {
      unsigned long insn = template[ix];
 
      /* Emit mov pc,rx if bx is not permitted.  */
      if (htab->fix_v4bx == 1 && (insn & 0x0ffffff0) == 0x012fff10)
	insn = (insn & 0xf000000f) | 0x01a0f000;
      put_arm_insn (htab, output_bfd, insn, (char *)contents + ix*4);
    }
}
 
/* Finish up the dynamic sections.  */
 
static bfd_boolean
elf32_arm_finish_dynamic_sections (bfd * output_bfd, struct bfd_link_info * info)
{
  bfd * dynobj;
  asection * sgot;
  asection * sdyn;
  struct elf32_arm_link_hash_table *htab;
 
  htab = elf32_arm_hash_table (info);
  if (htab == NULL)
    return FALSE;
 
  dynobj = elf_hash_table (info)->dynobj;
 
  sgot = htab->root.sgotplt;
  sdyn = bfd_get_section_by_name (dynobj, ".dynamic");
 
  if (elf_hash_table (info)->dynamic_sections_created)
    {
      asection *splt;
      Elf32_External_Dyn *dyncon, *dynconend;
 
      splt = htab->root.splt;
      BFD_ASSERT (splt != NULL && sdyn != NULL);
      BFD_ASSERT (htab->symbian_p || sgot != NULL);
 
      dyncon = (Elf32_External_Dyn *) sdyn->contents;
      dynconend = (Elf32_External_Dyn *) (sdyn->contents + sdyn->size);
 
      for (; dyncon < dynconend; dyncon++)
	{
	  Elf_Internal_Dyn dyn;
	  const char * name;
	  asection * s;
 
	  bfd_elf32_swap_dyn_in (dynobj, dyncon, &dyn);
 
	  switch (dyn.d_tag)
	    {
	      unsigned int type;
 
	    default:
	      if (htab->vxworks_p
		  && elf_vxworks_finish_dynamic_entry (output_bfd, &dyn))
		bfd_elf32_swap_dyn_out (output_bfd, &dyn, dyncon);
	      break;
 
	    case DT_HASH:
	      name = ".hash";
	      goto get_vma_if_bpabi;
	    case DT_STRTAB:
	      name = ".dynstr";
	      goto get_vma_if_bpabi;
	    case DT_SYMTAB:
	      name = ".dynsym";
	      goto get_vma_if_bpabi;
	    case DT_VERSYM:
	      name = ".gnu.version";
	      goto get_vma_if_bpabi;
	    case DT_VERDEF:
	      name = ".gnu.version_d";
	      goto get_vma_if_bpabi;
	    case DT_VERNEED:
	      name = ".gnu.version_r";
	      goto get_vma_if_bpabi;
 
	    case DT_PLTGOT:
	      name = ".got";
	      goto get_vma;
	    case DT_JMPREL:
	      name = RELOC_SECTION (htab, ".plt");
	    get_vma:
	      s = bfd_get_section_by_name (output_bfd, name);
	      BFD_ASSERT (s != NULL);
	      if (!htab->symbian_p)
		dyn.d_un.d_ptr = s->vma;
	      else
		/* In the BPABI, tags in the PT_DYNAMIC section point
		   at the file offset, not the memory address, for the
		   convenience of the post linker.  */
		dyn.d_un.d_ptr = s->filepos;
	      bfd_elf32_swap_dyn_out (output_bfd, &dyn, dyncon);
	      break;
 
	    get_vma_if_bpabi:
	      if (htab->symbian_p)
		goto get_vma;
	      break;
 
	    case DT_PLTRELSZ:
	      s = htab->root.srelplt;
	      BFD_ASSERT (s != NULL);
	      dyn.d_un.d_val = s->size;
	      bfd_elf32_swap_dyn_out (output_bfd, &dyn, dyncon);
	      break;
 
	    case DT_RELSZ:
	    case DT_RELASZ:
	      if (!htab->symbian_p)
		{
		  /* My reading of the SVR4 ABI indicates that the
		     procedure linkage table relocs (DT_JMPREL) should be
		     included in the overall relocs (DT_REL).  This is
		     what Solaris does.  However, UnixWare can not handle
		     that case.  Therefore, we override the DT_RELSZ entry
		     here to make it not include the JMPREL relocs.  Since
		     the linker script arranges for .rel(a).plt to follow all
		     other relocation sections, we don't have to worry
		     about changing the DT_REL entry.  */
		  s = htab->root.srelplt;
		  if (s != NULL)
		    dyn.d_un.d_val -= s->size;
		  bfd_elf32_swap_dyn_out (output_bfd, &dyn, dyncon);
		  break;
		}
	      /* Fall through.  */
 
	    case DT_REL:
	    case DT_RELA:
	      /* In the BPABI, the DT_REL tag must point at the file
		 offset, not the VMA, of the first relocation
		 section.  So, we use code similar to that in
		 elflink.c, but do not check for SHF_ALLOC on the
		 relcoation section, since relocations sections are
		 never allocated under the BPABI.  The comments above
		 about Unixware notwithstanding, we include all of the
		 relocations here.  */
	      if (htab->symbian_p)
		{
		  unsigned int i;
		  type = ((dyn.d_tag == DT_REL || dyn.d_tag == DT_RELSZ)
			  ? SHT_REL : SHT_RELA);
		  dyn.d_un.d_val = 0;
		  for (i = 1; i < elf_numsections (output_bfd); i++)
		    {
		      Elf_Internal_Shdr *hdr
			= elf_elfsections (output_bfd)[i];
		      if (hdr->sh_type == type)
			{
			  if (dyn.d_tag == DT_RELSZ
			      || dyn.d_tag == DT_RELASZ)
			    dyn.d_un.d_val += hdr->sh_size;
			  else if ((ufile_ptr) hdr->sh_offset
				   <= dyn.d_un.d_val - 1)
			    dyn.d_un.d_val = hdr->sh_offset;
			}
		    }
		  bfd_elf32_swap_dyn_out (output_bfd, &dyn, dyncon);
		}
	      break;
 
	    case DT_TLSDESC_PLT:
              s = htab->root.splt;
	      dyn.d_un.d_ptr = (s->output_section->vma + s->output_offset
				+ htab->dt_tlsdesc_plt);
	      bfd_elf32_swap_dyn_out (output_bfd, &dyn, dyncon);
	      break;
 
	    case DT_TLSDESC_GOT:
              s = htab->root.sgot;
	      dyn.d_un.d_ptr = (s->output_section->vma + s->output_offset
		      		+ htab->dt_tlsdesc_got);
	      bfd_elf32_swap_dyn_out (output_bfd, &dyn, dyncon);
	      break;
 
	      /* Set the bottom bit of DT_INIT/FINI if the
		 corresponding function is Thumb.  */
	    case DT_INIT:
	      name = info->init_function;
	      goto get_sym;
	    case DT_FINI:
	      name = info->fini_function;
	    get_sym:
	      /* If it wasn't set by elf_bfd_final_link
		 then there is nothing to adjust.  */
	      if (dyn.d_un.d_val != 0)
		{
		  struct elf_link_hash_entry * eh;
 
		  eh = elf_link_hash_lookup (elf_hash_table (info), name,
					     FALSE, FALSE, TRUE);
		  if (eh != NULL && eh->target_internal == ST_BRANCH_TO_THUMB)
		    {
		      dyn.d_un.d_val |= 1;
		      bfd_elf32_swap_dyn_out (output_bfd, &dyn, dyncon);
		    }
		}
	      break;
	    }
	}
 
      /* Fill in the first entry in the procedure linkage table.  */
      if (splt->size > 0 && htab->plt_header_size)
	{
	  const bfd_vma *plt0_entry;
	  bfd_vma got_address, plt_address, got_displacement;
 
	  /* Calculate the addresses of the GOT and PLT.  */
	  got_address = sgot->output_section->vma + sgot->output_offset;
	  plt_address = splt->output_section->vma + splt->output_offset;
 
	  if (htab->vxworks_p)
	    {
	      /* The VxWorks GOT is relocated by the dynamic linker.
		 Therefore, we must emit relocations rather than simply
		 computing the values now.  */
	      Elf_Internal_Rela rel;
 
	      plt0_entry = elf32_arm_vxworks_exec_plt0_entry;
	      put_arm_insn (htab, output_bfd, plt0_entry[0],
			    splt->contents + 0);
	      put_arm_insn (htab, output_bfd, plt0_entry[1],
			    splt->contents + 4);
	      put_arm_insn (htab, output_bfd, plt0_entry[2],
			    splt->contents + 8);
	      bfd_put_32 (output_bfd, got_address, splt->contents + 12);
 
	      /* Generate a relocation for _GLOBAL_OFFSET_TABLE_.  */
	      rel.r_offset = plt_address + 12;
	      rel.r_info = ELF32_R_INFO (htab->root.hgot->indx, R_ARM_ABS32);
	      rel.r_addend = 0;
	      SWAP_RELOC_OUT (htab) (output_bfd, &rel,
				     htab->srelplt2->contents);
	    }
	  else
	    {
	      got_displacement = got_address - (plt_address + 16);
 
	      plt0_entry = elf32_arm_plt0_entry;
	      put_arm_insn (htab, output_bfd, plt0_entry[0],
			    splt->contents + 0);
	      put_arm_insn (htab, output_bfd, plt0_entry[1],
			    splt->contents + 4);
	      put_arm_insn (htab, output_bfd, plt0_entry[2],
			    splt->contents + 8);
	      put_arm_insn (htab, output_bfd, plt0_entry[3],
			    splt->contents + 12);
 
#ifdef FOUR_WORD_PLT
	      /* The displacement value goes in the otherwise-unused
		 last word of the second entry.  */
	      bfd_put_32 (output_bfd, got_displacement, splt->contents + 28);
#else
	      bfd_put_32 (output_bfd, got_displacement, splt->contents + 16);
#endif
	    }
	}
 
      /* UnixWare sets the entsize of .plt to 4, although that doesn't
	 really seem like the right value.  */
      if (splt->output_section->owner == output_bfd)
	elf_section_data (splt->output_section)->this_hdr.sh_entsize = 4;
 
      if (htab->dt_tlsdesc_plt)
	{
	  bfd_vma got_address
	    = sgot->output_section->vma + sgot->output_offset;
	  bfd_vma gotplt_address = (htab->root.sgot->output_section->vma
				    + htab->root.sgot->output_offset);
	  bfd_vma plt_address
	    = splt->output_section->vma + splt->output_offset;
 
	  arm_put_trampoline (htab, output_bfd, 
			      splt->contents + htab->dt_tlsdesc_plt,
			      dl_tlsdesc_lazy_trampoline, 6);
 
	  bfd_put_32 (output_bfd,
		      gotplt_address + htab->dt_tlsdesc_got
		      - (plt_address + htab->dt_tlsdesc_plt)
		      - dl_tlsdesc_lazy_trampoline[6],
		      splt->contents + htab->dt_tlsdesc_plt + 24);
	  bfd_put_32 (output_bfd,
		      got_address - (plt_address + htab->dt_tlsdesc_plt)
		      - dl_tlsdesc_lazy_trampoline[7],
		      splt->contents + htab->dt_tlsdesc_plt + 24 + 4);
	}
 
      if (htab->tls_trampoline)
	{
	  arm_put_trampoline (htab, output_bfd, 
			      splt->contents + htab->tls_trampoline,
			      tls_trampoline, 3);
#ifdef FOUR_WORD_PLT
	  bfd_put_32 (output_bfd, 0x00000000,
		      splt->contents + htab->tls_trampoline + 12);
#endif 
	}
 
      if (htab->vxworks_p && !info->shared && htab->root.splt->size > 0)
	{
	  /* Correct the .rel(a).plt.unloaded relocations.  They will have
	     incorrect symbol indexes.  */
	  int num_plts;
	  unsigned char *p;
 
	  num_plts = ((htab->root.splt->size - htab->plt_header_size)
		      / htab->plt_entry_size);
	  p = htab->srelplt2->contents + RELOC_SIZE (htab);
 
	  for (; num_plts; num_plts--)
	    {
	      Elf_Internal_Rela rel;
 
	      SWAP_RELOC_IN (htab) (output_bfd, p, &rel);
	      rel.r_info = ELF32_R_INFO (htab->root.hgot->indx, R_ARM_ABS32);
	      SWAP_RELOC_OUT (htab) (output_bfd, &rel, p);
	      p += RELOC_SIZE (htab);
 
	      SWAP_RELOC_IN (htab) (output_bfd, p, &rel);
	      rel.r_info = ELF32_R_INFO (htab->root.hplt->indx, R_ARM_ABS32);
	      SWAP_RELOC_OUT (htab) (output_bfd, &rel, p);
	      p += RELOC_SIZE (htab);
	    }
	}
    }
 
  /* Fill in the first three entries in the global offset table.  */
  if (sgot)
    {
      if (sgot->size > 0)
	{
	  if (sdyn == NULL)
	    bfd_put_32 (output_bfd, (bfd_vma) 0, sgot->contents);
	  else
	    bfd_put_32 (output_bfd,
			sdyn->output_section->vma + sdyn->output_offset,
			sgot->contents);
	  bfd_put_32 (output_bfd, (bfd_vma) 0, sgot->contents + 4);
	  bfd_put_32 (output_bfd, (bfd_vma) 0, sgot->contents + 8);
	}
 
      elf_section_data (sgot->output_section)->this_hdr.sh_entsize = 4;
    }
 
  return TRUE;
}
 
static void
elf32_arm_post_process_headers (bfd * abfd, struct bfd_link_info * link_info ATTRIBUTE_UNUSED)
{
  Elf_Internal_Ehdr * i_ehdrp;	/* ELF file header, internal form.  */
  struct elf32_arm_link_hash_table *globals;
 
  i_ehdrp = elf_elfheader (abfd);
 
  if (EF_ARM_EABI_VERSION (i_ehdrp->e_flags) == EF_ARM_EABI_UNKNOWN)
    i_ehdrp->e_ident[EI_OSABI] = ELFOSABI_ARM;
  else
    i_ehdrp->e_ident[EI_OSABI] = 0;
  i_ehdrp->e_ident[EI_ABIVERSION] = ARM_ELF_ABI_VERSION;
 
  if (link_info)
    {
      globals = elf32_arm_hash_table (link_info);
      if (globals != NULL && globals->byteswap_code)
	i_ehdrp->e_flags |= EF_ARM_BE8;
    }
}
 
static enum elf_reloc_type_class
elf32_arm_reloc_type_class (const Elf_Internal_Rela *rela)
{
  switch ((int) ELF32_R_TYPE (rela->r_info))
    {
    case R_ARM_RELATIVE:
      return reloc_class_relative;
    case R_ARM_JUMP_SLOT:
      return reloc_class_plt;
    case R_ARM_COPY:
      return reloc_class_copy;
    default:
      return reloc_class_normal;
    }
}
 
/* Set the right machine number for an Arm ELF file.  */
 
static bfd_boolean
elf32_arm_section_flags (flagword *flags, const Elf_Internal_Shdr *hdr)
{
  if (hdr->sh_type == SHT_NOTE)
    *flags |= SEC_LINK_ONCE | SEC_LINK_DUPLICATES_SAME_CONTENTS;
 
  return TRUE;
}
 
static void
elf32_arm_final_write_processing (bfd *abfd, bfd_boolean linker ATTRIBUTE_UNUSED)
{
  bfd_arm_update_notes (abfd, ARM_NOTE_SECTION);
}
 
/* Return TRUE if this is an unwinding table entry.  */
 
static bfd_boolean
is_arm_elf_unwind_section_name (bfd * abfd ATTRIBUTE_UNUSED, const char * name)
{
  return (CONST_STRNEQ (name, ELF_STRING_ARM_unwind)
	  || CONST_STRNEQ (name, ELF_STRING_ARM_unwind_once));
}
 
 
/* Set the type and flags for an ARM section.  We do this by
   the section name, which is a hack, but ought to work.  */
 
static bfd_boolean
elf32_arm_fake_sections (bfd * abfd, Elf_Internal_Shdr * hdr, asection * sec)
{
  const char * name;
 
  name = bfd_get_section_name (abfd, sec);
 
  if (is_arm_elf_unwind_section_name (abfd, name))
    {
      hdr->sh_type = SHT_ARM_EXIDX;
      hdr->sh_flags |= SHF_LINK_ORDER;
    }
  return TRUE;
}
 
/* Handle an ARM specific section when reading an object file.  This is
   called when bfd_section_from_shdr finds a section with an unknown
   type.  */
 
static bfd_boolean
elf32_arm_section_from_shdr (bfd *abfd,
			     Elf_Internal_Shdr * hdr,
			     const char *name,
			     int shindex)
{
  /* There ought to be a place to keep ELF backend specific flags, but
     at the moment there isn't one.  We just keep track of the
     sections by their name, instead.  Fortunately, the ABI gives
     names for all the ARM specific sections, so we will probably get
     away with this.  */
  switch (hdr->sh_type)
    {
    case SHT_ARM_EXIDX:
    case SHT_ARM_PREEMPTMAP:
    case SHT_ARM_ATTRIBUTES:
      break;
 
    default:
      return FALSE;
    }
 
  if (! _bfd_elf_make_section_from_shdr (abfd, hdr, name, shindex))
    return FALSE;
 
  return TRUE;
}
 
static _arm_elf_section_data *
get_arm_elf_section_data (asection * sec)
{
  if (sec && sec->owner && is_arm_elf (sec->owner))
    return elf32_arm_section_data (sec);
  else
    return NULL;
}
 
typedef struct
{
  void *finfo;
  struct bfd_link_info *info;
  asection *sec;
  int sec_shndx;
  int (*func) (void *, const char *, Elf_Internal_Sym *,
	       asection *, struct elf_link_hash_entry *);
} output_arch_syminfo;
 
enum map_symbol_type
{
  ARM_MAP_ARM,
  ARM_MAP_THUMB,
  ARM_MAP_DATA
};
 
 
/* Output a single mapping symbol.  */
 
static bfd_boolean
elf32_arm_output_map_sym (output_arch_syminfo *osi,
			  enum map_symbol_type type,
			  bfd_vma offset)
{
  static const char *names[3] = {"$a", "$t", "$d"};
  Elf_Internal_Sym sym;
 
  sym.st_value = osi->sec->output_section->vma
		 + osi->sec->output_offset
		 + offset;
  sym.st_size = 0;
  sym.st_other = 0;
  sym.st_info = ELF_ST_INFO (STB_LOCAL, STT_NOTYPE);
  sym.st_shndx = osi->sec_shndx;
  sym.st_target_internal = 0;
  elf32_arm_section_map_add (osi->sec, names[type][1], offset);
  return osi->func (osi->finfo, names[type], &sym, osi->sec, NULL) == 1;
}
 
/* Output mapping symbols for the PLT entry described by ROOT_PLT and ARM_PLT.
   IS_IPLT_ENTRY_P says whether the PLT is in .iplt rather than .plt.  */
 
static bfd_boolean
elf32_arm_output_plt_map_1 (output_arch_syminfo *osi,
			    bfd_boolean is_iplt_entry_p,
			    union gotplt_union *root_plt,
			    struct arm_plt_info *arm_plt)
{
  struct elf32_arm_link_hash_table *htab;
  bfd_vma addr, plt_header_size;
 
  if (root_plt->offset == (bfd_vma) -1)
    return TRUE;
 
  htab = elf32_arm_hash_table (osi->info);
  if (htab == NULL)
    return FALSE;
 
  if (is_iplt_entry_p)
    {
      osi->sec = htab->root.iplt;
      plt_header_size = 0;
    }
  else
    {
      osi->sec = htab->root.splt;
      plt_header_size = htab->plt_header_size;
    }
  osi->sec_shndx = (_bfd_elf_section_from_bfd_section
		    (osi->info->output_bfd, osi->sec->output_section));
 
  addr = root_plt->offset & -2;
  if (htab->symbian_p)
    {
      if (!elf32_arm_output_map_sym (osi, ARM_MAP_ARM, addr))
	return FALSE;
      if (!elf32_arm_output_map_sym (osi, ARM_MAP_DATA, addr + 4))
	return FALSE;
    }
  else if (htab->vxworks_p)
    {
      if (!elf32_arm_output_map_sym (osi, ARM_MAP_ARM, addr))
	return FALSE;
      if (!elf32_arm_output_map_sym (osi, ARM_MAP_DATA, addr + 8))
	return FALSE;
      if (!elf32_arm_output_map_sym (osi, ARM_MAP_ARM, addr + 12))
	return FALSE;
      if (!elf32_arm_output_map_sym (osi, ARM_MAP_DATA, addr + 20))
	return FALSE;
    }
  else
    {
      bfd_boolean thumb_stub_p;
 
      thumb_stub_p = elf32_arm_plt_needs_thumb_stub_p (osi->info, arm_plt);
      if (thumb_stub_p)
	{
	  if (!elf32_arm_output_map_sym (osi, ARM_MAP_THUMB, addr - 4))
	    return FALSE;
	}
#ifdef FOUR_WORD_PLT
      if (!elf32_arm_output_map_sym (osi, ARM_MAP_ARM, addr))
	return FALSE;
      if (!elf32_arm_output_map_sym (osi, ARM_MAP_DATA, addr + 12))
	return FALSE;
#else
      /* A three-word PLT with no Thumb thunk contains only Arm code,
	 so only need to output a mapping symbol for the first PLT entry and
	 entries with thumb thunks.  */
      if (thumb_stub_p || addr == plt_header_size)
	{
	  if (!elf32_arm_output_map_sym (osi, ARM_MAP_ARM, addr))
	    return FALSE;
	}
#endif
    }
 
  return TRUE;
}
 
/* Output mapping symbols for PLT entries associated with H.  */
 
static bfd_boolean
elf32_arm_output_plt_map (struct elf_link_hash_entry *h, void *inf)
{
  output_arch_syminfo *osi = (output_arch_syminfo *) inf;
  struct elf32_arm_link_hash_entry *eh;
 
  if (h->root.type == bfd_link_hash_indirect)
    return TRUE;
 
  if (h->root.type == bfd_link_hash_warning)
    /* When warning symbols are created, they **replace** the "real"
       entry in the hash table, thus we never get to see the real
       symbol in a hash traversal.  So look at it now.  */
    h = (struct elf_link_hash_entry *) h->root.u.i.link;
 
  eh = (struct elf32_arm_link_hash_entry *) h;
  return elf32_arm_output_plt_map_1 (osi, SYMBOL_CALLS_LOCAL (osi->info, h),
				     &h->plt, &eh->plt);
}
 
/* Output a single local symbol for a generated stub.  */
 
static bfd_boolean
elf32_arm_output_stub_sym (output_arch_syminfo *osi, const char *name,
			   bfd_vma offset, bfd_vma size)
{
  Elf_Internal_Sym sym;
 
  sym.st_value = osi->sec->output_section->vma
		 + osi->sec->output_offset
		 + offset;
  sym.st_size = size;
  sym.st_other = 0;
  sym.st_info = ELF_ST_INFO (STB_LOCAL, STT_FUNC);
  sym.st_shndx = osi->sec_shndx;
  sym.st_target_internal = 0;
  return osi->func (osi->finfo, name, &sym, osi->sec, NULL) == 1;
}
 
static bfd_boolean
arm_map_one_stub (struct bfd_hash_entry * gen_entry,
		  void * in_arg)
{
  struct elf32_arm_stub_hash_entry *stub_entry;
  asection *stub_sec;
  bfd_vma addr;
  char *stub_name;
  output_arch_syminfo *osi;
  const insn_sequence *template_sequence;
  enum stub_insn_type prev_type;
  int size;
  int i;
  enum map_symbol_type sym_type;
 
  /* Massage our args to the form they really have.  */
  stub_entry = (struct elf32_arm_stub_hash_entry *) gen_entry;
  osi = (output_arch_syminfo *) in_arg;
 
  stub_sec = stub_entry->stub_sec;
 
  /* Ensure this stub is attached to the current section being
     processed.  */
  if (stub_sec != osi->sec)
    return TRUE;
 
  addr = (bfd_vma) stub_entry->stub_offset;
  stub_name = stub_entry->output_name;
 
  template_sequence = stub_entry->stub_template;
  switch (template_sequence[0].type)
    {
    case ARM_TYPE:
      if (!elf32_arm_output_stub_sym (osi, stub_name, addr, stub_entry->stub_size))
	return FALSE;
      break;
    case THUMB16_TYPE:
    case THUMB32_TYPE:
      if (!elf32_arm_output_stub_sym (osi, stub_name, addr | 1,
				      stub_entry->stub_size))
	return FALSE;
      break;
    default:
      BFD_FAIL ();
      return 0;
    }
 
  prev_type = DATA_TYPE;
  size = 0;
  for (i = 0; i < stub_entry->stub_template_size; i++)
    {
      switch (template_sequence[i].type)
	{
	case ARM_TYPE:
	  sym_type = ARM_MAP_ARM;
	  break;
 
	case THUMB16_TYPE:
	case THUMB32_TYPE:
	  sym_type = ARM_MAP_THUMB;
	  break;
 
	case DATA_TYPE:
	  sym_type = ARM_MAP_DATA;
	  break;
 
	default:
	  BFD_FAIL ();
	  return FALSE;
	}
 
      if (template_sequence[i].type != prev_type)
	{
	  prev_type = template_sequence[i].type;
	  if (!elf32_arm_output_map_sym (osi, sym_type, addr + size))
	    return FALSE;
	}
 
      switch (template_sequence[i].type)
	{
	case ARM_TYPE:
	case THUMB32_TYPE:
	  size += 4;
	  break;
 
	case THUMB16_TYPE:
	  size += 2;
	  break;
 
	case DATA_TYPE:
	  size += 4;
	  break;
 
	default:
	  BFD_FAIL ();
	  return FALSE;
	}
    }
 
  return TRUE;
}
 
/* Output mapping symbols for linker generated sections,
   and for those data-only sections that do not have a
   $d.  */
 
static bfd_boolean
elf32_arm_output_arch_local_syms (bfd *output_bfd,
				  struct bfd_link_info *info,
				  void *finfo,
				  int (*func) (void *, const char *,
					       Elf_Internal_Sym *,
					       asection *,
					       struct elf_link_hash_entry *))
{
  output_arch_syminfo osi;
  struct elf32_arm_link_hash_table *htab;
  bfd_vma offset;
  bfd_size_type size;
  bfd *input_bfd;
 
  htab = elf32_arm_hash_table (info);
  if (htab == NULL)
    return FALSE;
 
  check_use_blx (htab);
 
  osi.finfo = finfo;
  osi.info = info;
  osi.func = func;
 
  /* Add a $d mapping symbol to data-only sections that
     don't have any mapping symbol.  This may result in (harmless) redundant
     mapping symbols.  */
  for (input_bfd = info->input_bfds;
       input_bfd != NULL;
       input_bfd = input_bfd->link_next)
    {
      if ((input_bfd->flags & (BFD_LINKER_CREATED | HAS_SYMS)) == HAS_SYMS)
	for (osi.sec = input_bfd->sections;
	     osi.sec != NULL;
	     osi.sec = osi.sec->next)
	  {
	    if (osi.sec->output_section != NULL
		&& ((osi.sec->output_section->flags & (SEC_ALLOC | SEC_CODE))
		    != 0)
		&& (osi.sec->flags & (SEC_HAS_CONTENTS | SEC_LINKER_CREATED))
		   == SEC_HAS_CONTENTS
		&& get_arm_elf_section_data (osi.sec) != NULL
		&& get_arm_elf_section_data (osi.sec)->mapcount == 0
		&& osi.sec->size > 0)
	      {
		osi.sec_shndx = _bfd_elf_section_from_bfd_section
		  (output_bfd, osi.sec->output_section);
		if (osi.sec_shndx != (int)SHN_BAD)
		  elf32_arm_output_map_sym (&osi, ARM_MAP_DATA, 0);
	      }
	  }
    }
 
  /* ARM->Thumb glue.  */
  if (htab->arm_glue_size > 0)
    {
      osi.sec = bfd_get_section_by_name (htab->bfd_of_glue_owner,
					 ARM2THUMB_GLUE_SECTION_NAME);
 
      osi.sec_shndx = _bfd_elf_section_from_bfd_section
	  (output_bfd, osi.sec->output_section);
      if (info->shared || htab->root.is_relocatable_executable
	  || htab->pic_veneer)
	size = ARM2THUMB_PIC_GLUE_SIZE;
      else if (htab->use_blx)
	size = ARM2THUMB_V5_STATIC_GLUE_SIZE;
      else
	size = ARM2THUMB_STATIC_GLUE_SIZE;
 
      for (offset = 0; offset < htab->arm_glue_size; offset += size)
	{
	  elf32_arm_output_map_sym (&osi, ARM_MAP_ARM, offset);
	  elf32_arm_output_map_sym (&osi, ARM_MAP_DATA, offset + size - 4);
	}
    }
 
  /* Thumb->ARM glue.  */
  if (htab->thumb_glue_size > 0)
    {
      osi.sec = bfd_get_section_by_name (htab->bfd_of_glue_owner,
					 THUMB2ARM_GLUE_SECTION_NAME);
 
      osi.sec_shndx = _bfd_elf_section_from_bfd_section
	  (output_bfd, osi.sec->output_section);
      size = THUMB2ARM_GLUE_SIZE;
 
      for (offset = 0; offset < htab->thumb_glue_size; offset += size)
	{
	  elf32_arm_output_map_sym (&osi, ARM_MAP_THUMB, offset);
	  elf32_arm_output_map_sym (&osi, ARM_MAP_ARM, offset + 4);
	}
    }
 
  /* ARMv4 BX veneers.  */
  if (htab->bx_glue_size > 0)
    {
      osi.sec = bfd_get_section_by_name (htab->bfd_of_glue_owner,
					 ARM_BX_GLUE_SECTION_NAME);
 
      osi.sec_shndx = _bfd_elf_section_from_bfd_section
	  (output_bfd, osi.sec->output_section);
 
      elf32_arm_output_map_sym (&osi, ARM_MAP_ARM, 0);
    }
 
  /* Long calls stubs.  */
  if (htab->stub_bfd && htab->stub_bfd->sections)
    {
      asection* stub_sec;
 
      for (stub_sec = htab->stub_bfd->sections;
	   stub_sec != NULL;
	   stub_sec = stub_sec->next)
	{
	  /* Ignore non-stub sections.  */
	  if (!strstr (stub_sec->name, STUB_SUFFIX))
	    continue;
 
	  osi.sec = stub_sec;
 
	  osi.sec_shndx = _bfd_elf_section_from_bfd_section
	    (output_bfd, osi.sec->output_section);
 
	  bfd_hash_traverse (&htab->stub_hash_table, arm_map_one_stub, &osi);
	}
    }
 
  /* Finally, output mapping symbols for the PLT.  */
  if (htab->root.splt && htab->root.splt->size > 0)
    {
      osi.sec = htab->root.splt;
      osi.sec_shndx = (_bfd_elf_section_from_bfd_section
		       (output_bfd, osi.sec->output_section));
 
      /* Output mapping symbols for the plt header.  SymbianOS does not have a
	 plt header.  */
      if (htab->vxworks_p)
	{
	  /* VxWorks shared libraries have no PLT header.  */
	  if (!info->shared)
	    {
	      if (!elf32_arm_output_map_sym (&osi, ARM_MAP_ARM, 0))
		return FALSE;
	      if (!elf32_arm_output_map_sym (&osi, ARM_MAP_DATA, 12))
		return FALSE;
	    }
	}
      else if (!htab->symbian_p)
	{
	  if (!elf32_arm_output_map_sym (&osi, ARM_MAP_ARM, 0))
	    return FALSE;
#ifndef FOUR_WORD_PLT
	  if (!elf32_arm_output_map_sym (&osi, ARM_MAP_DATA, 16))
	    return FALSE;
#endif
	}
    }
  if ((htab->root.splt && htab->root.splt->size > 0)
      || (htab->root.iplt && htab->root.iplt->size > 0))
    {
      elf_link_hash_traverse (&htab->root, elf32_arm_output_plt_map, &osi);
      for (input_bfd = info->input_bfds;
	   input_bfd != NULL;
	   input_bfd = input_bfd->link_next)
	{
	  struct arm_local_iplt_info **local_iplt;
	  unsigned int i, num_syms;
 
	  local_iplt = elf32_arm_local_iplt (input_bfd);
	  if (local_iplt != NULL)
	    {
	      num_syms = elf_symtab_hdr (input_bfd).sh_info;
	      for (i = 0; i < num_syms; i++)
		if (local_iplt[i] != NULL
		    && !elf32_arm_output_plt_map_1 (&osi, TRUE,
						    &local_iplt[i]->root,
						    &local_iplt[i]->arm))
		  return FALSE;
	    }
	}
    }
  if (htab->dt_tlsdesc_plt != 0)
    {
      /* Mapping symbols for the lazy tls trampoline.  */
      if (!elf32_arm_output_map_sym (&osi, ARM_MAP_ARM, htab->dt_tlsdesc_plt))
	return FALSE;
 
      if (!elf32_arm_output_map_sym (&osi, ARM_MAP_DATA,
				     htab->dt_tlsdesc_plt + 24))
	return FALSE;
    }
  if (htab->tls_trampoline != 0)
    {
      /* Mapping symbols for the tls trampoline.  */
      if (!elf32_arm_output_map_sym (&osi, ARM_MAP_ARM, htab->tls_trampoline))
	return FALSE;
#ifdef FOUR_WORD_PLT
      if (!elf32_arm_output_map_sym (&osi, ARM_MAP_DATA,
				     htab->tls_trampoline + 12))
	return FALSE;
#endif 
    }
 
  return TRUE;
}
 
/* Allocate target specific section data.  */
 
static bfd_boolean
elf32_arm_new_section_hook (bfd *abfd, asection *sec)
{
  if (!sec->used_by_bfd)
    {
      _arm_elf_section_data *sdata;
      bfd_size_type amt = sizeof (*sdata);
 
      sdata = (_arm_elf_section_data *) bfd_zalloc (abfd, amt);
      if (sdata == NULL)
	return FALSE;
      sec->used_by_bfd = sdata;
    }
 
  return _bfd_elf_new_section_hook (abfd, sec);
}
 
 
/* Used to order a list of mapping symbols by address.  */
 
static int
elf32_arm_compare_mapping (const void * a, const void * b)
{
  const elf32_arm_section_map *amap = (const elf32_arm_section_map *) a;
  const elf32_arm_section_map *bmap = (const elf32_arm_section_map *) b;
 
  if (amap->vma > bmap->vma)
    return 1;
  else if (amap->vma < bmap->vma)
    return -1;
  else if (amap->type > bmap->type)
    /* Ensure results do not depend on the host qsort for objects with
       multiple mapping symbols at the same address by sorting on type
       after vma.  */
    return 1;
  else if (amap->type < bmap->type)
    return -1;
  else
    return 0;
}
 
/* Add OFFSET to lower 31 bits of ADDR, leaving other bits unmodified.  */
 
static unsigned long
offset_prel31 (unsigned long addr, bfd_vma offset)
{
  return (addr & ~0x7ffffffful) | ((addr + offset) & 0x7ffffffful);
}
 
/* Copy an .ARM.exidx table entry, adding OFFSET to (applied) PREL31
   relocations.  */
 
static void
copy_exidx_entry (bfd *output_bfd, bfd_byte *to, bfd_byte *from, bfd_vma offset)
{
  unsigned long first_word = bfd_get_32 (output_bfd, from);
  unsigned long second_word = bfd_get_32 (output_bfd, from + 4);
 
  /* High bit of first word is supposed to be zero.  */
  if ((first_word & 0x80000000ul) == 0)
    first_word = offset_prel31 (first_word, offset);
 
  /* If the high bit of the first word is clear, and the bit pattern is not 0x1
     (EXIDX_CANTUNWIND), this is an offset to an .ARM.extab entry.  */
  if ((second_word != 0x1) && ((second_word & 0x80000000ul) == 0))
    second_word = offset_prel31 (second_word, offset);
 
  bfd_put_32 (output_bfd, first_word, to);
  bfd_put_32 (output_bfd, second_word, to + 4);
}
 
/* Data for make_branch_to_a8_stub().  */
 
struct a8_branch_to_stub_data {
  asection *writing_section;
  bfd_byte *contents;
};
 
 
/* Helper to insert branches to Cortex-A8 erratum stubs in the right
   places for a particular section.  */
 
static bfd_boolean
make_branch_to_a8_stub (struct bfd_hash_entry *gen_entry,
                       void *in_arg)
{
  struct elf32_arm_stub_hash_entry *stub_entry;
  struct a8_branch_to_stub_data *data;
  bfd_byte *contents;
  unsigned long branch_insn;
  bfd_vma veneered_insn_loc, veneer_entry_loc;
  bfd_signed_vma branch_offset;
  bfd *abfd;
  unsigned int target;
 
  stub_entry = (struct elf32_arm_stub_hash_entry *) gen_entry;
  data = (struct a8_branch_to_stub_data *) in_arg;
 
  if (stub_entry->target_section != data->writing_section
      || stub_entry->stub_type < arm_stub_a8_veneer_lwm)
    return TRUE;
 
  contents = data->contents;
 
  veneered_insn_loc = stub_entry->target_section->output_section->vma
		      + stub_entry->target_section->output_offset
		      + stub_entry->target_value;
 
  veneer_entry_loc = stub_entry->stub_sec->output_section->vma
		     + stub_entry->stub_sec->output_offset
		     + stub_entry->stub_offset;
 
  if (stub_entry->stub_type == arm_stub_a8_veneer_blx)
    veneered_insn_loc &= ~3u;
 
  branch_offset = veneer_entry_loc - veneered_insn_loc - 4;
 
  abfd = stub_entry->target_section->owner;
  target = stub_entry->target_value;
 
  /* We attempt to avoid this condition by setting stubs_always_after_branch
     in elf32_arm_size_stubs if we've enabled the Cortex-A8 erratum workaround.
     This check is just to be on the safe side...  */
  if ((veneered_insn_loc & ~0xfff) == (veneer_entry_loc & ~0xfff))
    {
      (*_bfd_error_handler) (_("%B: error: Cortex-A8 erratum stub is "
			       "allocated in unsafe location"), abfd);
      return FALSE;
    }
 
  switch (stub_entry->stub_type)
    {
    case arm_stub_a8_veneer_b:
    case arm_stub_a8_veneer_b_cond:
      branch_insn = 0xf0009000;
      goto jump24;
 
    case arm_stub_a8_veneer_blx:
      branch_insn = 0xf000e800;
      goto jump24;
 
    case arm_stub_a8_veneer_bl:
      {
	unsigned int i1, j1, i2, j2, s;
 
	branch_insn = 0xf000d000;
 
      jump24:
	if (branch_offset < -16777216 || branch_offset > 16777214)
	  {
	    /* There's not much we can do apart from complain if this
	       happens.  */
	    (*_bfd_error_handler) (_("%B: error: Cortex-A8 erratum stub out "
				     "of range (input file too large)"), abfd);
	    return FALSE;
	  }
 
	/* i1 = not(j1 eor s), so:
	   not i1 = j1 eor s
	   j1 = (not i1) eor s.  */
 
	branch_insn |= (branch_offset >> 1) & 0x7ff;
	branch_insn |= ((branch_offset >> 12) & 0x3ff) << 16;
	i2 = (branch_offset >> 22) & 1;
	i1 = (branch_offset >> 23) & 1;
	s = (branch_offset >> 24) & 1;
	j1 = (!i1) ^ s;
	j2 = (!i2) ^ s;
	branch_insn |= j2 << 11;
	branch_insn |= j1 << 13;
	branch_insn |= s << 26;
      }
      break;
 
    default:
      BFD_FAIL ();
      return FALSE;
    }
 
  bfd_put_16 (abfd, (branch_insn >> 16) & 0xffff, &contents[target]);
  bfd_put_16 (abfd, branch_insn & 0xffff, &contents[target + 2]);
 
  return TRUE;
}
 
/* Do code byteswapping.  Return FALSE afterwards so that the section is
   written out as normal.  */
 
static bfd_boolean
elf32_arm_write_section (bfd *output_bfd,
			 struct bfd_link_info *link_info,
			 asection *sec,
			 bfd_byte *contents)
{
  unsigned int mapcount, errcount;
  _arm_elf_section_data *arm_data;
  struct elf32_arm_link_hash_table *globals = elf32_arm_hash_table (link_info);
  elf32_arm_section_map *map;
  elf32_vfp11_erratum_list *errnode;
  bfd_vma ptr;
  bfd_vma end;
  bfd_vma offset = sec->output_section->vma + sec->output_offset;
  bfd_byte tmp;
  unsigned int i;
 
  if (globals == NULL)
    return FALSE;
 
  /* If this section has not been allocated an _arm_elf_section_data
     structure then we cannot record anything.  */
  arm_data = get_arm_elf_section_data (sec);
  if (arm_data == NULL)
    return FALSE;
 
  mapcount = arm_data->mapcount;
  map = arm_data->map;
  errcount = arm_data->erratumcount;
 
  if (errcount != 0)
    {
      unsigned int endianflip = bfd_big_endian (output_bfd) ? 3 : 0;
 
      for (errnode = arm_data->erratumlist; errnode != 0;
           errnode = errnode->next)
        {
          bfd_vma target = errnode->vma - offset;
 
          switch (errnode->type)
            {
            case VFP11_ERRATUM_BRANCH_TO_ARM_VENEER:
              {
                bfd_vma branch_to_veneer;
                /* Original condition code of instruction, plus bit mask for
                   ARM B instruction.  */
                unsigned int insn = (errnode->u.b.vfp_insn & 0xf0000000)
                                  | 0x0a000000;
 
		/* The instruction is before the label.  */
		target -= 4;
 
		/* Above offset included in -4 below.  */
		branch_to_veneer = errnode->u.b.veneer->vma
                                   - errnode->vma - 4;
 
		if ((signed) branch_to_veneer < -(1 << 25)
		    || (signed) branch_to_veneer >= (1 << 25))
		  (*_bfd_error_handler) (_("%B: error: VFP11 veneer out of "
					   "range"), output_bfd);
 
                insn |= (branch_to_veneer >> 2) & 0xffffff;
                contents[endianflip ^ target] = insn & 0xff;
                contents[endianflip ^ (target + 1)] = (insn >> 8) & 0xff;
                contents[endianflip ^ (target + 2)] = (insn >> 16) & 0xff;
                contents[endianflip ^ (target + 3)] = (insn >> 24) & 0xff;
              }
              break;
 
	    case VFP11_ERRATUM_ARM_VENEER:
              {
                bfd_vma branch_from_veneer;
                unsigned int insn;
 
                /* Take size of veneer into account.  */
                branch_from_veneer = errnode->u.v.branch->vma
                                     - errnode->vma - 12;
 
		if ((signed) branch_from_veneer < -(1 << 25)
		    || (signed) branch_from_veneer >= (1 << 25))
		  (*_bfd_error_handler) (_("%B: error: VFP11 veneer out of "
					   "range"), output_bfd);
 
                /* Original instruction.  */
                insn = errnode->u.v.branch->u.b.vfp_insn;
                contents[endianflip ^ target] = insn & 0xff;
                contents[endianflip ^ (target + 1)] = (insn >> 8) & 0xff;
                contents[endianflip ^ (target + 2)] = (insn >> 16) & 0xff;
                contents[endianflip ^ (target + 3)] = (insn >> 24) & 0xff;
 
                /* Branch back to insn after original insn.  */
                insn = 0xea000000 | ((branch_from_veneer >> 2) & 0xffffff);
                contents[endianflip ^ (target + 4)] = insn & 0xff;
                contents[endianflip ^ (target + 5)] = (insn >> 8) & 0xff;
                contents[endianflip ^ (target + 6)] = (insn >> 16) & 0xff;
                contents[endianflip ^ (target + 7)] = (insn >> 24) & 0xff;
              }
              break;
 
            default:
              abort ();
            }
        }
    }
 
  if (arm_data->elf.this_hdr.sh_type == SHT_ARM_EXIDX)
    {
      arm_unwind_table_edit *edit_node
        = arm_data->u.exidx.unwind_edit_list;
      /* Now, sec->size is the size of the section we will write.  The original
         size (before we merged duplicate entries and inserted EXIDX_CANTUNWIND
	 markers) was sec->rawsize.  (This isn't the case if we perform no
	 edits, then rawsize will be zero and we should use size).  */
      bfd_byte *edited_contents = (bfd_byte *) bfd_malloc (sec->size);
      unsigned int input_size = sec->rawsize ? sec->rawsize : sec->size;
      unsigned int in_index, out_index;
      bfd_vma add_to_offsets = 0;
 
      for (in_index = 0, out_index = 0; in_index * 8 < input_size || edit_node;)
        {
	  if (edit_node)
	    {
	      unsigned int edit_index = edit_node->index;
 
	      if (in_index < edit_index && in_index * 8 < input_size)
	        {
		  copy_exidx_entry (output_bfd, edited_contents + out_index * 8,
				    contents + in_index * 8, add_to_offsets);
		  out_index++;
		  in_index++;
		}
	      else if (in_index == edit_index
		       || (in_index * 8 >= input_size
			   && edit_index == UINT_MAX))
	        {
		  switch (edit_node->type)
		    {
		    case DELETE_EXIDX_ENTRY:
		      in_index++;
		      add_to_offsets += 8;
		      break;
 
		    case INSERT_EXIDX_CANTUNWIND_AT_END:
		      {
		        asection *text_sec = edit_node->linked_section;
			bfd_vma text_offset = text_sec->output_section->vma
					      + text_sec->output_offset
					      + text_sec->size;
			bfd_vma exidx_offset = offset + out_index * 8;
		        unsigned long prel31_offset;
 
			/* Note: this is meant to be equivalent to an
			   R_ARM_PREL31 relocation.  These synthetic
			   EXIDX_CANTUNWIND markers are not relocated by the
			   usual BFD method.  */
			prel31_offset = (text_offset - exidx_offset)
					& 0x7ffffffful;
 
			/* First address we can't unwind.  */
			bfd_put_32 (output_bfd, prel31_offset,
				    &edited_contents[out_index * 8]);
 
			/* Code for EXIDX_CANTUNWIND.  */
			bfd_put_32 (output_bfd, 0x1,
				    &edited_contents[out_index * 8 + 4]);
 
			out_index++;
			add_to_offsets -= 8;
		      }
		      break;
		    }
 
		  edit_node = edit_node->next;
		}
	    }
	  else
	    {
	      /* No more edits, copy remaining entries verbatim.  */
	      copy_exidx_entry (output_bfd, edited_contents + out_index * 8,
				contents + in_index * 8, add_to_offsets);
	      out_index++;
	      in_index++;
	    }
	}
 
      if (!(sec->flags & SEC_EXCLUDE) && !(sec->flags & SEC_NEVER_LOAD))
	bfd_set_section_contents (output_bfd, sec->output_section,
				  edited_contents,
				  (file_ptr) sec->output_offset, sec->size);
 
      return TRUE;
    }
 
  /* Fix code to point to Cortex-A8 erratum stubs.  */
  if (globals->fix_cortex_a8)
    {
      struct a8_branch_to_stub_data data;
 
      data.writing_section = sec;
      data.contents = contents;
 
      bfd_hash_traverse (&globals->stub_hash_table, make_branch_to_a8_stub,
			 &data);
    }
 
  if (mapcount == 0)
    return FALSE;
 
  if (globals->byteswap_code)
    {
      qsort (map, mapcount, sizeof (* map), elf32_arm_compare_mapping);
 
      ptr = map[0].vma;
      for (i = 0; i < mapcount; i++)
        {
          if (i == mapcount - 1)
	    end = sec->size;
          else
            end = map[i + 1].vma;
 
          switch (map[i].type)
	    {
	    case 'a':
	      /* Byte swap code words.  */
	      while (ptr + 3 < end)
	        {
	          tmp = contents[ptr];
	          contents[ptr] = contents[ptr + 3];
	          contents[ptr + 3] = tmp;
	          tmp = contents[ptr + 1];
	          contents[ptr + 1] = contents[ptr + 2];
	          contents[ptr + 2] = tmp;
	          ptr += 4;
	        }
	      break;
 
	    case 't':
	      /* Byte swap code halfwords.  */
	      while (ptr + 1 < end)
	        {
	          tmp = contents[ptr];
	          contents[ptr] = contents[ptr + 1];
	          contents[ptr + 1] = tmp;
	          ptr += 2;
	        }
	      break;
 
	    case 'd':
	      /* Leave data alone.  */
	      break;
	    }
          ptr = end;
        }
    }
 
  free (map);
  arm_data->mapcount = -1;
  arm_data->mapsize = 0;
  arm_data->map = NULL;
 
  return FALSE;
}
 
/* Mangle thumb function symbols as we read them in.  */
 
static bfd_boolean
elf32_arm_swap_symbol_in (bfd * abfd,
			  const void *psrc,
			  const void *pshn,
			  Elf_Internal_Sym *dst)
{
  if (!bfd_elf32_swap_symbol_in (abfd, psrc, pshn, dst))
    return FALSE;
 
  /* New EABI objects mark thumb function symbols by setting the low bit of
     the address.  */
  if (ELF_ST_TYPE (dst->st_info) == STT_FUNC
      || ELF_ST_TYPE (dst->st_info) == STT_GNU_IFUNC)
    {
      if (dst->st_value & 1)
	{
	  dst->st_value &= ~(bfd_vma) 1;
	  dst->st_target_internal = ST_BRANCH_TO_THUMB;
	}
      else
	dst->st_target_internal = ST_BRANCH_TO_ARM;
    }
  else if (ELF_ST_TYPE (dst->st_info) == STT_ARM_TFUNC)
    {
      dst->st_info = ELF_ST_INFO (ELF_ST_BIND (dst->st_info), STT_FUNC);
      dst->st_target_internal = ST_BRANCH_TO_THUMB;
    }
  else if (ELF_ST_TYPE (dst->st_info) == STT_SECTION)
    dst->st_target_internal = ST_BRANCH_LONG;
  else
    dst->st_target_internal = ST_BRANCH_UNKNOWN;
 
  return TRUE;
}
 
 
/* Mangle thumb function symbols as we write them out.  */
 
static void
elf32_arm_swap_symbol_out (bfd *abfd,
			   const Elf_Internal_Sym *src,
			   void *cdst,
			   void *shndx)
{
  Elf_Internal_Sym newsym;
 
  /* We convert STT_ARM_TFUNC symbols into STT_FUNC with the low bit
     of the address set, as per the new EABI.  We do this unconditionally
     because objcopy does not set the elf header flags until after
     it writes out the symbol table.  */
  if (src->st_target_internal == ST_BRANCH_TO_THUMB)
    {
      newsym = *src;
      if (ELF_ST_TYPE (src->st_info) != STT_GNU_IFUNC)
	newsym.st_info = ELF_ST_INFO (ELF_ST_BIND (src->st_info), STT_FUNC);
      if (newsym.st_shndx != SHN_UNDEF)
        {
          /* Do this only for defined symbols. At link type, the static
             linker will simulate the work of dynamic linker of resolving
             symbols and will carry over the thumbness of found symbols to
             the output symbol table. It's not clear how it happens, but
             the thumbness of undefined symbols can well be different at
             runtime, and writing '1' for them will be confusing for users
             and possibly for dynamic linker itself.
          */
          newsym.st_value |= 1;
        }
 
      src = &newsym;
    }
  bfd_elf32_swap_symbol_out (abfd, src, cdst, shndx);
}
 
/* Add the PT_ARM_EXIDX program header.  */
 
static bfd_boolean
elf32_arm_modify_segment_map (bfd *abfd,
			      struct bfd_link_info *info ATTRIBUTE_UNUSED)
{
  struct elf_segment_map *m;
  asection *sec;
 
  sec = bfd_get_section_by_name (abfd, ".ARM.exidx");
  if (sec != NULL && (sec->flags & SEC_LOAD) != 0)
    {
      /* If there is already a PT_ARM_EXIDX header, then we do not
	 want to add another one.  This situation arises when running
	 "strip"; the input binary already has the header.  */
      m = elf_tdata (abfd)->segment_map;
      while (m && m->p_type != PT_ARM_EXIDX)
	m = m->next;
      if (!m)
	{
	  m = (struct elf_segment_map *)
              bfd_zalloc (abfd, sizeof (struct elf_segment_map));
	  if (m == NULL)
	    return FALSE;
	  m->p_type = PT_ARM_EXIDX;
	  m->count = 1;
	  m->sections[0] = sec;
 
	  m->next = elf_tdata (abfd)->segment_map;
	  elf_tdata (abfd)->segment_map = m;
	}
    }
 
  return TRUE;
}
 
/* We may add a PT_ARM_EXIDX program header.  */
 
static int
elf32_arm_additional_program_headers (bfd *abfd,
				      struct bfd_link_info *info ATTRIBUTE_UNUSED)
{
  asection *sec;
 
  sec = bfd_get_section_by_name (abfd, ".ARM.exidx");
  if (sec != NULL && (sec->flags & SEC_LOAD) != 0)
    return 1;
  else
    return 0;
}
 
/* Hook called by the linker routine which adds symbols from an object
   file.  */
 
static bfd_boolean
elf32_arm_add_symbol_hook (bfd *abfd, struct bfd_link_info *info,
			   Elf_Internal_Sym *sym, const char **namep,
			   flagword *flagsp, asection **secp, bfd_vma *valp)
{
  if ((abfd->flags & DYNAMIC) == 0
      && (ELF_ST_TYPE (sym->st_info) == STT_GNU_IFUNC
	  || ELF_ST_BIND (sym->st_info) == STB_GNU_UNIQUE))
    elf_tdata (info->output_bfd)->has_gnu_symbols = TRUE;
 
  if (elf32_arm_hash_table (info)->vxworks_p
      && !elf_vxworks_add_symbol_hook (abfd, info, sym, namep,
				       flagsp, secp, valp))
    return FALSE;
 
  return TRUE;
}
 
/* We use this to override swap_symbol_in and swap_symbol_out.  */
const struct elf_size_info elf32_arm_size_info =
{
  sizeof (Elf32_External_Ehdr),
  sizeof (Elf32_External_Phdr),
  sizeof (Elf32_External_Shdr),
  sizeof (Elf32_External_Rel),
  sizeof (Elf32_External_Rela),
  sizeof (Elf32_External_Sym),
  sizeof (Elf32_External_Dyn),
  sizeof (Elf_External_Note),
  4,
  1,
  32, 2,
  ELFCLASS32, EV_CURRENT,
  bfd_elf32_write_out_phdrs,
  bfd_elf32_write_shdrs_and_ehdr,
  bfd_elf32_checksum_contents,
  bfd_elf32_write_relocs,
  elf32_arm_swap_symbol_in,
  elf32_arm_swap_symbol_out,
  bfd_elf32_slurp_reloc_table,
  bfd_elf32_slurp_symbol_table,
  bfd_elf32_swap_dyn_in,
  bfd_elf32_swap_dyn_out,
  bfd_elf32_swap_reloc_in,
  bfd_elf32_swap_reloc_out,
  bfd_elf32_swap_reloca_in,
  bfd_elf32_swap_reloca_out
};
 
#define ELF_ARCH			bfd_arch_arm
#define ELF_TARGET_ID			ARM_ELF_DATA
#define ELF_MACHINE_CODE		EM_ARM
#ifdef __QNXTARGET__
#define ELF_MAXPAGESIZE			0x1000
#else
#define ELF_MAXPAGESIZE			0x8000
#endif
#define ELF_MINPAGESIZE			0x1000
#define ELF_COMMONPAGESIZE		0x1000
 
#define bfd_elf32_mkobject		        elf32_arm_mkobject
 
#define bfd_elf32_bfd_copy_private_bfd_data	elf32_arm_copy_private_bfd_data
#define bfd_elf32_bfd_merge_private_bfd_data	elf32_arm_merge_private_bfd_data
#define bfd_elf32_bfd_set_private_flags		elf32_arm_set_private_flags
#define bfd_elf32_bfd_print_private_bfd_data	elf32_arm_print_private_bfd_data
#define bfd_elf32_bfd_link_hash_table_create    elf32_arm_link_hash_table_create
#define bfd_elf32_bfd_link_hash_table_free      elf32_arm_hash_table_free
#define bfd_elf32_bfd_reloc_type_lookup		elf32_arm_reloc_type_lookup
#define bfd_elf32_bfd_reloc_name_lookup	elf32_arm_reloc_name_lookup
#define bfd_elf32_find_nearest_line	        elf32_arm_find_nearest_line
#define bfd_elf32_find_inliner_info	        elf32_arm_find_inliner_info
#define bfd_elf32_new_section_hook		elf32_arm_new_section_hook
#define bfd_elf32_bfd_is_target_special_symbol	elf32_arm_is_target_special_symbol
#define bfd_elf32_bfd_final_link		elf32_arm_final_link
 
#define elf_backend_get_symbol_type             elf32_arm_get_symbol_type
#define elf_backend_gc_mark_hook                elf32_arm_gc_mark_hook
#define elf_backend_gc_mark_extra_sections	elf32_arm_gc_mark_extra_sections
#define elf_backend_gc_sweep_hook               elf32_arm_gc_sweep_hook
#define elf_backend_check_relocs                elf32_arm_check_relocs
#define elf_backend_relocate_section		elf32_arm_relocate_section
#define elf_backend_write_section		elf32_arm_write_section
#define elf_backend_adjust_dynamic_symbol	elf32_arm_adjust_dynamic_symbol
#define elf_backend_create_dynamic_sections     elf32_arm_create_dynamic_sections
#define elf_backend_finish_dynamic_symbol	elf32_arm_finish_dynamic_symbol
#define elf_backend_finish_dynamic_sections	elf32_arm_finish_dynamic_sections
#define elf_backend_size_dynamic_sections	elf32_arm_size_dynamic_sections
#define elf_backend_always_size_sections	elf32_arm_always_size_sections
#define elf_backend_init_index_section		_bfd_elf_init_2_index_sections
#define elf_backend_post_process_headers	elf32_arm_post_process_headers
#define elf_backend_reloc_type_class		elf32_arm_reloc_type_class
#define elf_backend_object_p			elf32_arm_object_p
#define elf_backend_section_flags		elf32_arm_section_flags
#define elf_backend_fake_sections  		elf32_arm_fake_sections
#define elf_backend_section_from_shdr  		elf32_arm_section_from_shdr
#define elf_backend_final_write_processing      elf32_arm_final_write_processing
#define elf_backend_copy_indirect_symbol        elf32_arm_copy_indirect_symbol
#define elf_backend_size_info			elf32_arm_size_info
#define elf_backend_modify_segment_map		elf32_arm_modify_segment_map
#define elf_backend_additional_program_headers  elf32_arm_additional_program_headers
#define elf_backend_output_arch_local_syms      elf32_arm_output_arch_local_syms
#define elf_backend_begin_write_processing      elf32_arm_begin_write_processing
#define elf_backend_add_symbol_hook		elf32_arm_add_symbol_hook
 
#define elf_backend_can_refcount       1
#define elf_backend_can_gc_sections    1
#define elf_backend_plt_readonly       1
#define elf_backend_want_got_plt       1
#define elf_backend_want_plt_sym       0
#define elf_backend_may_use_rel_p      1
#define elf_backend_may_use_rela_p     0
#define elf_backend_default_use_rela_p 0
 
#define elf_backend_got_header_size	12
 
#undef  elf_backend_obj_attrs_vendor
#define elf_backend_obj_attrs_vendor		"aeabi"
#undef  elf_backend_obj_attrs_section
#define elf_backend_obj_attrs_section		".ARM.attributes"
#undef  elf_backend_obj_attrs_arg_type
#define elf_backend_obj_attrs_arg_type		elf32_arm_obj_attrs_arg_type
#undef  elf_backend_obj_attrs_section_type
#define elf_backend_obj_attrs_section_type	SHT_ARM_ATTRIBUTES
#define elf_backend_obj_attrs_order	elf32_arm_obj_attrs_order
#define elf_backend_obj_attrs_handle_unknown elf32_arm_obj_attrs_handle_unknown
 
#include "elf32-target.h"
 
/* VxWorks Targets.  */
 
#undef  TARGET_LITTLE_SYM
#define TARGET_LITTLE_SYM               bfd_elf32_littlearm_vxworks_vec
#undef  TARGET_LITTLE_NAME
#define TARGET_LITTLE_NAME              "elf32-littlearm-vxworks"
#undef  TARGET_BIG_SYM
#define TARGET_BIG_SYM                  bfd_elf32_bigarm_vxworks_vec
#undef  TARGET_BIG_NAME
#define TARGET_BIG_NAME                 "elf32-bigarm-vxworks"
 
/* Like elf32_arm_link_hash_table_create -- but overrides
   appropriately for VxWorks.  */
 
static struct bfd_link_hash_table *
elf32_arm_vxworks_link_hash_table_create (bfd *abfd)
{
  struct bfd_link_hash_table *ret;
 
  ret = elf32_arm_link_hash_table_create (abfd);
  if (ret)
    {
      struct elf32_arm_link_hash_table *htab
	= (struct elf32_arm_link_hash_table *) ret;
      htab->use_rel = 0;
      htab->vxworks_p = 1;
    }
  return ret;
}
 
static void
elf32_arm_vxworks_final_write_processing (bfd *abfd, bfd_boolean linker)
{
  elf32_arm_final_write_processing (abfd, linker);
  elf_vxworks_final_write_processing (abfd, linker);
}
 
#undef  elf32_bed
#define elf32_bed elf32_arm_vxworks_bed
 
#undef  bfd_elf32_bfd_link_hash_table_create
#define bfd_elf32_bfd_link_hash_table_create	elf32_arm_vxworks_link_hash_table_create
#undef  elf_backend_final_write_processing
#define elf_backend_final_write_processing	elf32_arm_vxworks_final_write_processing
#undef  elf_backend_emit_relocs
#define elf_backend_emit_relocs			elf_vxworks_emit_relocs
 
#undef  elf_backend_may_use_rel_p
#define elf_backend_may_use_rel_p	0
#undef  elf_backend_may_use_rela_p
#define elf_backend_may_use_rela_p	1
#undef  elf_backend_default_use_rela_p
#define elf_backend_default_use_rela_p	1
#undef  elf_backend_want_plt_sym
#define elf_backend_want_plt_sym	1
#undef  ELF_MAXPAGESIZE
#define ELF_MAXPAGESIZE			0x1000
 
#include "elf32-target.h"
 
 
/* Merge backend specific data from an object file to the output
   object file when linking.  */
 
static bfd_boolean
elf32_arm_merge_private_bfd_data (bfd * ibfd, bfd * obfd)
{
  flagword out_flags;
  flagword in_flags;
  bfd_boolean flags_compatible = TRUE;
  asection *sec;
 
  /* Check if we have the same endianness.  */
  if (! _bfd_generic_verify_endian_match (ibfd, obfd))
    return FALSE;
 
  if (! is_arm_elf (ibfd) || ! is_arm_elf (obfd))
    return TRUE;
 
  if (!elf32_arm_merge_eabi_attributes (ibfd, obfd))
    return FALSE;
 
  /* The input BFD must have had its flags initialised.  */
  /* The following seems bogus to me -- The flags are initialized in
     the assembler but I don't think an elf_flags_init field is
     written into the object.  */
  /* BFD_ASSERT (elf_flags_init (ibfd)); */
 
  in_flags  = elf_elfheader (ibfd)->e_flags;
  out_flags = elf_elfheader (obfd)->e_flags;
 
  /* In theory there is no reason why we couldn't handle this.  However
     in practice it isn't even close to working and there is no real
     reason to want it.  */
  if (EF_ARM_EABI_VERSION (in_flags) >= EF_ARM_EABI_VER4
      && !(ibfd->flags & DYNAMIC)
      && (in_flags & EF_ARM_BE8))
    {
      _bfd_error_handler (_("error: %B is already in final BE8 format"),
			  ibfd);
      return FALSE;
    }
 
  if (!elf_flags_init (obfd))
    {
      /* If the input is the default architecture and had the default
	 flags then do not bother setting the flags for the output
	 architecture, instead allow future merges to do this.  If no
	 future merges ever set these flags then they will retain their
         uninitialised values, which surprise surprise, correspond
         to the default values.  */
      if (bfd_get_arch_info (ibfd)->the_default
	  && elf_elfheader (ibfd)->e_flags == 0)
	return TRUE;
 
      elf_flags_init (obfd) = TRUE;
      elf_elfheader (obfd)->e_flags = in_flags;
 
      if (bfd_get_arch (obfd) == bfd_get_arch (ibfd)
	  && bfd_get_arch_info (obfd)->the_default)
	return bfd_set_arch_mach (obfd, bfd_get_arch (ibfd), bfd_get_mach (ibfd));
 
      return TRUE;
    }
 
  /* Determine what should happen if the input ARM architecture
     does not match the output ARM architecture.  */
  if (! bfd_arm_merge_machines (ibfd, obfd))
    return FALSE;
 
  /* Identical flags must be compatible.  */
  if (in_flags == out_flags)
    return TRUE;
 
  /* Check to see if the input BFD actually contains any sections.  If
     not, its flags may not have been initialised either, but it
     cannot actually cause any incompatiblity.  Do not short-circuit
     dynamic objects; their section list may be emptied by
    elf_link_add_object_symbols.
 
    Also check to see if there are no code sections in the input.
    In this case there is no need to check for code specific flags.
    XXX - do we need to worry about floating-point format compatability
    in data sections ?  */
  if (!(ibfd->flags & DYNAMIC))
    {
      bfd_boolean null_input_bfd = TRUE;
      bfd_boolean only_data_sections = TRUE;
 
      for (sec = ibfd->sections; sec != NULL; sec = sec->next)
	{
	  /* Ignore synthetic glue sections.  */
	  if (strcmp (sec->name, ".glue_7")
	      && strcmp (sec->name, ".glue_7t"))
	    {
	      if ((bfd_get_section_flags (ibfd, sec)
		   & (SEC_LOAD | SEC_CODE | SEC_HAS_CONTENTS))
		  == (SEC_LOAD | SEC_CODE | SEC_HAS_CONTENTS))
	    	only_data_sections = FALSE;
 
	      null_input_bfd = FALSE;
	      break;
	    }
	}
 
      if (null_input_bfd || only_data_sections)
	return TRUE;
    }
 
  /* Complain about various flag mismatches.  */
  if (!elf32_arm_versions_compatible (EF_ARM_EABI_VERSION (in_flags),
				      EF_ARM_EABI_VERSION (out_flags)))
    {
      _bfd_error_handler
	(_("error: Source object %B has EABI version %d, but target %B has EABI version %d"),
	 ibfd, obfd,
	 (in_flags & EF_ARM_EABIMASK) >> 24,
	 (out_flags & EF_ARM_EABIMASK) >> 24);
      return FALSE;
    }
 
  /* Not sure what needs to be checked for EABI versions >= 1.  */
  /* VxWorks libraries do not use these flags.  */
  if (get_elf_backend_data (obfd) != &elf32_arm_vxworks_bed
      && get_elf_backend_data (ibfd) != &elf32_arm_vxworks_bed
      && EF_ARM_EABI_VERSION (in_flags) == EF_ARM_EABI_UNKNOWN)
    {
      if ((in_flags & EF_ARM_APCS_26) != (out_flags & EF_ARM_APCS_26))
	{
	  _bfd_error_handler
	    (_("error: %B is compiled for APCS-%d, whereas target %B uses APCS-%d"),
	     ibfd, obfd,
	     in_flags & EF_ARM_APCS_26 ? 26 : 32,
	     out_flags & EF_ARM_APCS_26 ? 26 : 32);
	  flags_compatible = FALSE;
	}
 
      if ((in_flags & EF_ARM_APCS_FLOAT) != (out_flags & EF_ARM_APCS_FLOAT))
	{
	  if (in_flags & EF_ARM_APCS_FLOAT)
	    _bfd_error_handler
	      (_("error: %B passes floats in float registers, whereas %B passes them in integer registers"),
	       ibfd, obfd);
	  else
	    _bfd_error_handler
	      (_("error: %B passes floats in integer registers, whereas %B passes them in float registers"),
	       ibfd, obfd);
 
	  flags_compatible = FALSE;
	}
 
      if ((in_flags & EF_ARM_VFP_FLOAT) != (out_flags & EF_ARM_VFP_FLOAT))
	{
	  if (in_flags & EF_ARM_VFP_FLOAT)
	    _bfd_error_handler
	      (_("error: %B uses VFP instructions, whereas %B does not"),
	       ibfd, obfd);
	  else
	    _bfd_error_handler
	      (_("error: %B uses FPA instructions, whereas %B does not"),
	       ibfd, obfd);
 
	  flags_compatible = FALSE;
	}
 
      if ((in_flags & EF_ARM_MAVERICK_FLOAT) != (out_flags & EF_ARM_MAVERICK_FLOAT))
	{
	  if (in_flags & EF_ARM_MAVERICK_FLOAT)
	    _bfd_error_handler
	      (_("error: %B uses Maverick instructions, whereas %B does not"),
	       ibfd, obfd);
	  else
	    _bfd_error_handler
	      (_("error: %B does not use Maverick instructions, whereas %B does"),
	       ibfd, obfd);
 
	  flags_compatible = FALSE;
	}
 
#ifdef EF_ARM_SOFT_FLOAT
      if ((in_flags & EF_ARM_SOFT_FLOAT) != (out_flags & EF_ARM_SOFT_FLOAT))
	{
	  /* We can allow interworking between code that is VFP format
	     layout, and uses either soft float or integer regs for
	     passing floating point arguments and results.  We already
	     know that the APCS_FLOAT flags match; similarly for VFP
	     flags.  */
	  if ((in_flags & EF_ARM_APCS_FLOAT) != 0
	      || (in_flags & EF_ARM_VFP_FLOAT) == 0)
	    {
	      if (in_flags & EF_ARM_SOFT_FLOAT)
		_bfd_error_handler
		  (_("error: %B uses software FP, whereas %B uses hardware FP"),
		   ibfd, obfd);
	      else
		_bfd_error_handler
		  (_("error: %B uses hardware FP, whereas %B uses software FP"),
		   ibfd, obfd);
 
	      flags_compatible = FALSE;
	    }
	}
#endif
 
      /* Interworking mismatch is only a warning.  */
      if ((in_flags & EF_ARM_INTERWORK) != (out_flags & EF_ARM_INTERWORK))
	{
	  if (in_flags & EF_ARM_INTERWORK)
	    {
	      _bfd_error_handler
		(_("Warning: %B supports interworking, whereas %B does not"),
		 ibfd, obfd);
	    }
	  else
	    {
	      _bfd_error_handler
		(_("Warning: %B does not support interworking, whereas %B does"),
		 ibfd, obfd);
	    }
	}
    }
 
  return flags_compatible;
}
 
 
/* Symbian OS Targets.  */
 
#undef  TARGET_LITTLE_SYM
#define TARGET_LITTLE_SYM               bfd_elf32_littlearm_symbian_vec
#undef  TARGET_LITTLE_NAME
#define TARGET_LITTLE_NAME              "elf32-littlearm-symbian"
#undef  TARGET_BIG_SYM
#define TARGET_BIG_SYM                  bfd_elf32_bigarm_symbian_vec
#undef  TARGET_BIG_NAME
#define TARGET_BIG_NAME                 "elf32-bigarm-symbian"
 
/* Like elf32_arm_link_hash_table_create -- but overrides
   appropriately for Symbian OS.  */
 
static struct bfd_link_hash_table *
elf32_arm_symbian_link_hash_table_create (bfd *abfd)
{
  struct bfd_link_hash_table *ret;
 
  ret = elf32_arm_link_hash_table_create (abfd);
  if (ret)
    {
      struct elf32_arm_link_hash_table *htab
	= (struct elf32_arm_link_hash_table *)ret;
      /* There is no PLT header for Symbian OS.  */
      htab->plt_header_size = 0;
      /* The PLT entries are each one instruction and one word.  */
      htab->plt_entry_size = 4 * ARRAY_SIZE (elf32_arm_symbian_plt_entry);
      htab->symbian_p = 1;
      /* Symbian uses armv5t or above, so use_blx is always true.  */
      htab->use_blx = 1;
      htab->root.is_relocatable_executable = 1;
    }
  return ret;
}
 
static const struct bfd_elf_special_section
elf32_arm_symbian_special_sections[] =
{
  /* In a BPABI executable, the dynamic linking sections do not go in
     the loadable read-only segment.  The post-linker may wish to
     refer to these sections, but they are not part of the final
     program image.  */
  { STRING_COMMA_LEN (".dynamic"),       0, SHT_DYNAMIC,  0 },
  { STRING_COMMA_LEN (".dynstr"),        0, SHT_STRTAB,   0 },
  { STRING_COMMA_LEN (".dynsym"),        0, SHT_DYNSYM,   0 },
  { STRING_COMMA_LEN (".got"),           0, SHT_PROGBITS, 0 },
  { STRING_COMMA_LEN (".hash"),          0, SHT_HASH,     0 },
  /* These sections do not need to be writable as the SymbianOS
     postlinker will arrange things so that no dynamic relocation is
     required.  */
  { STRING_COMMA_LEN (".init_array"),    0, SHT_INIT_ARRAY,    SHF_ALLOC },
  { STRING_COMMA_LEN (".fini_array"),    0, SHT_FINI_ARRAY,    SHF_ALLOC },
  { STRING_COMMA_LEN (".preinit_array"), 0, SHT_PREINIT_ARRAY, SHF_ALLOC },
  { NULL,                             0, 0, 0,                 0 }
};
 
static void
elf32_arm_symbian_begin_write_processing (bfd *abfd,
					  struct bfd_link_info *link_info)
{
  /* BPABI objects are never loaded directly by an OS kernel; they are
     processed by a postlinker first, into an OS-specific format.  If
     the D_PAGED bit is set on the file, BFD will align segments on
     page boundaries, so that an OS can directly map the file.  With
     BPABI objects, that just results in wasted space.  In addition,
     because we clear the D_PAGED bit, map_sections_to_segments will
     recognize that the program headers should not be mapped into any
     loadable segment.  */
  abfd->flags &= ~D_PAGED;
  elf32_arm_begin_write_processing (abfd, link_info);
}
 
static bfd_boolean
elf32_arm_symbian_modify_segment_map (bfd *abfd,
				      struct bfd_link_info *info)
{
  struct elf_segment_map *m;
  asection *dynsec;
 
  /* BPABI shared libraries and executables should have a PT_DYNAMIC
     segment.  However, because the .dynamic section is not marked
     with SEC_LOAD, the generic ELF code will not create such a
     segment.  */
  dynsec = bfd_get_section_by_name (abfd, ".dynamic");
  if (dynsec)
    {
      for (m = elf_tdata (abfd)->segment_map; m != NULL; m = m->next)
	if (m->p_type == PT_DYNAMIC)
	  break;
 
      if (m == NULL)
	{
	  m = _bfd_elf_make_dynamic_segment (abfd, dynsec);
	  m->next = elf_tdata (abfd)->segment_map;
	  elf_tdata (abfd)->segment_map = m;
	}
    }
 
  /* Also call the generic arm routine.  */
  return elf32_arm_modify_segment_map (abfd, info);
}
 
/* Return address for Ith PLT stub in section PLT, for relocation REL
   or (bfd_vma) -1 if it should not be included.  */
 
static bfd_vma
elf32_arm_symbian_plt_sym_val (bfd_vma i, const asection *plt,
			       const arelent *rel ATTRIBUTE_UNUSED)
{
  return plt->vma + 4 * ARRAY_SIZE (elf32_arm_symbian_plt_entry) * i;
}
 
 
#undef  elf32_bed
#define elf32_bed elf32_arm_symbian_bed
 
/* The dynamic sections are not allocated on SymbianOS; the postlinker
   will process them and then discard them.  */
#undef  ELF_DYNAMIC_SEC_FLAGS
#define ELF_DYNAMIC_SEC_FLAGS \
  (SEC_HAS_CONTENTS | SEC_IN_MEMORY | SEC_LINKER_CREATED)
 
#undef elf_backend_emit_relocs
 
#undef  bfd_elf32_bfd_link_hash_table_create
#define bfd_elf32_bfd_link_hash_table_create	elf32_arm_symbian_link_hash_table_create
#undef  elf_backend_special_sections
#define elf_backend_special_sections 		elf32_arm_symbian_special_sections
#undef  elf_backend_begin_write_processing
#define elf_backend_begin_write_processing	elf32_arm_symbian_begin_write_processing
#undef  elf_backend_final_write_processing
#define elf_backend_final_write_processing	elf32_arm_final_write_processing
 
#undef  elf_backend_modify_segment_map
#define elf_backend_modify_segment_map elf32_arm_symbian_modify_segment_map
 
/* There is no .got section for BPABI objects, and hence no header.  */
#undef  elf_backend_got_header_size
#define elf_backend_got_header_size 0
 
/* Similarly, there is no .got.plt section.  */
#undef  elf_backend_want_got_plt
#define elf_backend_want_got_plt 0
 
#undef  elf_backend_plt_sym_val
#define elf_backend_plt_sym_val		elf32_arm_symbian_plt_sym_val
 
#undef  elf_backend_may_use_rel_p
#define elf_backend_may_use_rel_p	1
#undef  elf_backend_may_use_rela_p
#define elf_backend_may_use_rela_p	0
#undef  elf_backend_default_use_rela_p
#define elf_backend_default_use_rela_p	0
#undef  elf_backend_want_plt_sym
#define elf_backend_want_plt_sym	0
#undef  ELF_MAXPAGESIZE
#define ELF_MAXPAGESIZE			0x8000
 
#include "elf32-target.h"
 

Go to most recent revision | Compare with Previous | Blame | View Log

powered by: WebSVN 2.1.0

© copyright 1999-2024 OpenCores.org, equivalent to Oliscience, all rights reserved. OpenCores®, registered trademark.